Add character/environment profiles, 3D conversion, archive, multi-character dialog

- Character profiles: extract from images/video or generate from text prompt;
  up to 6 profiles per image/video generation via IP-Adapter conditioning
- Environment profiles: same multi-slot system for scene/background conditioning
- Multi-character dialog for video: per-character TTS synthesis with voice profile
  support, sequential or manual timing, ffmpeg amix, Wav2Lip/SadTalker lip sync
- 2D↔3D conversion: image/video to stereo/anaglyph/depth/mesh and back;
  text/image to 3D model (GLB); turntable video rendering
- Generation archive: auto-save all outputs with configurable retention and
  hourly cleanup; browse/delete from web UI and API
- Voice profile enhancements: extract from video, PATCH update, GET by name;
  TTS endpoint now accepts voice_profile for F5-TTS cloning
- Image generation progress tracking via GET /v1/images/progress
- Web Studio: Profiles tab (Characters, Environments, Voices) with full managers;
  multi-slot character/environment selectors in all panels; character dialog
  section in all video panels; archive settings in Settings page
- Fix: Profiles tab sub-tab switching now correctly updates the content panel
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

codai/admin/routes.py

@@ -1556,6 +1556,11 @@ async def api_get_settings(username: str = Depends(require_admin)):
             "device_id": c.vulkan.device_id,
             "single_gpu": c.vulkan.single_gpu,
         },
+        "archive": {
+            "enabled": c.archive.enabled,
+            "directory": c.archive.directory,
+            "retention": c.archive.retention,
+        },
         "system_prompt": c.system_prompt,
         "tools_closer_prompt": c.tools_closer_prompt,
         "grammar_guided": c.grammar_guided,
@@ -1632,8 +1637,95 @@ async def api_save_settings(request: Request, username: str = Depends(require_ad
     if "parser" in data:
         c.parser = data["parser"]
 
+    if "archive" in data:
+        import os as _os
+        from codai.api.archive import archive_manager, RETENTION_OPTIONS
+        arc = data["archive"]
+        c.archive.enabled = bool(arc.get("enabled", c.archive.enabled))
+        raw_dir = (arc.get("directory") or "").strip()
+        c.archive.directory = raw_dir  # store as-is (empty = default)
+        ret = arc.get("retention", c.archive.retention)
+        c.archive.retention = ret if ret in RETENTION_OPTIONS else "never"
+        # Resolve for live reconfiguration
+        cfg_dir = str(config_manager.config_dir)
+        resolved = raw_dir if raw_dir and _os.path.isabs(raw_dir) else _os.path.join(cfg_dir, raw_dir or "archive")
+        archive_manager.configure(c.archive.enabled, resolved, c.archive.retention)
+
     config_manager.save_config()
     return {"success": True}
+
+
+# =============================================================================
+# Archive management
+# =============================================================================
+
+@router.get("/admin/archive", response_class=HTMLResponse)
+async def archive_page(request: Request, username: str = Depends(require_admin)):
+    return templates.TemplateResponse(request, "archive.html", {"username": username, "is_admin": True})
+
+
+@router.get("/admin/api/archive")
+async def api_archive_list(
+    limit: int = 50,
+    offset: int = 0,
+    username: str = Depends(require_admin),
+):
+    from codai.api.archive import archive_manager
+    entries = archive_manager.list_entries(limit=limit, offset=offset)
+    total = archive_manager.count_entries()
+    return {"entries": entries, "total": total}
+
+
+@router.get("/admin/api/archive/{gen_id}")
+async def api_archive_get(gen_id: str, username: str = Depends(require_admin)):
+    from codai.api.archive import archive_manager
+    entry = archive_manager.get_entry(gen_id)
+    if entry is None:
+        raise HTTPException(status_code=404, detail="Entry not found")
+    return entry
+
+
+@router.delete("/admin/api/archive/{gen_id}")
+async def api_archive_delete(gen_id: str, username: str = Depends(require_admin)):
+    from codai.api.archive import archive_manager
+    if not archive_manager.delete_entry(gen_id):
+        raise HTTPException(status_code=404, detail="Entry not found")
+    return {"success": True}
+
+
+@router.get("/admin/api/archive/{gen_id}/files/{filename}")
+async def api_archive_file(
+    gen_id: str,
+    filename: str,
+    username: str = Depends(require_auth),
+):
+    from fastapi.responses import FileResponse
+    from codai.api.archive import archive_manager
+    path = archive_manager.get_file_path(gen_id, filename)
+    if path is None:
+        raise HTTPException(status_code=404, detail="File not found")
+    import mimetypes
+    media_type = mimetypes.guess_type(filename)[0] or "application/octet-stream"
+    return FileResponse(path, media_type=media_type)
+
+
+@router.get("/admin/api/archive-settings")
+async def api_archive_settings_get(username: str = Depends(require_admin)):
+    if config_manager is None or config_manager.config is None:
+        raise HTTPException(status_code=503, detail="Config not ready")
+    import os as _os
+    c = config_manager.config.archive
+    from codai.api.archive import RETENTION_OPTIONS
+    default_dir = _os.path.join(str(config_manager.config_dir), "archive")
+    return {
+        "enabled": c.enabled,
+        "directory": c.directory,
+        "default_directory": default_dir,
+        "retention": c.retention,
+        "retention_options": RETENTION_OPTIONS,
+    }
+
+
 # --- HuggingFace model search proxy ---
 
 import re as _re
@@ -1794,6 +1886,110 @@ async def api_hf_model_files(model_id: str, username: str = Depends(require_admi
     return files
 
 
+# =============================================================================
+# Character profile management proxy (admin UI)
+# =============================================================================
+
+@router.get("/admin/api/characters")
+async def api_list_characters(username: str = Depends(require_auth)):
+    from codai.api.characters import _list_characters
+    return {"characters": _list_characters()}
+
+
+@router.get("/admin/api/characters/{name}")
+async def api_get_character(name: str, username: str = Depends(require_auth)):
+    from codai.api.characters import _load_character_meta, _load_character_images
+    meta = _load_character_meta(name)
+    if not meta:
+        raise HTTPException(status_code=404, detail=f"Character '{name}' not found")
+    images = _load_character_images(name)
+    return {
+        "name": meta['name'],
+        "description": meta.get('description', ''),
+        "image_count": meta['image_count'],
+        "created_at": meta['created_at'],
+        "images": [img.model_dump() for img in images],
+    }
+
+
+@router.delete("/admin/api/characters/{name}")
+async def api_delete_character(name: str, username: str = Depends(require_auth)):
+    import os as _os, shutil
+    from codai.api.characters import _char_dir
+    cdir = _char_dir(name)
+    if not _os.path.isdir(cdir):
+        raise HTTPException(status_code=404, detail=f"Character '{name}' not found")
+    shutil.rmtree(cdir)
+    return {"ok": True, "name": name}
+
+
+# =============================================================================
+# Environment profile management proxy (admin UI)
+# =============================================================================
+
+@router.get("/admin/api/environments")
+async def api_list_environments(username: str = Depends(require_auth)):
+    from codai.api.environments import _list_environments
+    return {"environments": _list_environments()}
+
+
+@router.get("/admin/api/environments/{name}")
+async def api_get_environment(name: str, username: str = Depends(require_auth)):
+    from codai.api.environments import _load_environment_meta, _load_environment_images
+    meta = _load_environment_meta(name)
+    if not meta:
+        raise HTTPException(status_code=404, detail=f"Environment '{name}' not found")
+    images = _load_environment_images(name)
+    return {
+        "name": meta['name'],
+        "description": meta.get('description', ''),
+        "image_count": meta['image_count'],
+        "created_at": meta['created_at'],
+        "images": [img.model_dump() for img in images],
+    }
+
+
+@router.delete("/admin/api/environments/{name}")
+async def api_delete_environment(name: str, username: str = Depends(require_auth)):
+    import os as _os, shutil
+    from codai.api.environments import _env_dir
+    edir = _env_dir(name)
+    if not _os.path.isdir(edir):
+        raise HTTPException(status_code=404, detail=f"Environment '{name}' not found")
+    shutil.rmtree(edir)
+    return {"ok": True, "name": name}
+
+
+# =============================================================================
+# Voice profile management proxy (admin UI)
+# =============================================================================
+
+@router.get("/admin/api/voices")
+async def api_list_voices(username: str = Depends(require_auth)):
+    from codai.api.voice_clone import _list_voices
+    return {"voices": _list_voices()}
+
+
+@router.get("/admin/api/voices/{name}")
+async def api_get_voice(name: str, username: str = Depends(require_auth)):
+    from codai.api.voice_clone import _load_voice
+    meta = _load_voice(name)
+    if not meta:
+        raise HTTPException(status_code=404, detail=f"Voice '{name}' not found")
+    return {"voice": meta}
+
+
+@router.delete("/admin/api/voices/{name}")
+async def api_delete_voice(name: str, username: str = Depends(require_auth)):
+    import os as _os, shutil
+    from codai.api.voice_clone import _voice_path
+    vdir = _voice_path(name)
+    if not _os.path.exists(vdir):
+        raise HTTPException(status_code=404, detail=f"Voice '{name}' not found")
+    shutil.rmtree(vdir)
+    return {"deleted": True, "name": name}
+
+
 @router.get("/admin/api/hf-model-info")
 async def api_hf_model_info(model_id: str, username: str = Depends(require_admin)):
     """Full metadata for a single HuggingFace model repo."""

codai/admin/templates/base.html

@@ -27,6 +27,7 @@
         <a href="/admin/models" class="nav-link {% if '/models' in request.url.path %}active{% endif %}">Models</a>
         <a href="/admin/tokens" class="nav-link {% if '/tokens' in request.url.path %}active{% endif %}">Tokens</a>
         <a href="/admin/users" class="nav-link {% if '/users' in request.url.path %}active{% endif %}">Users</a>
+        <a href="/admin/archive" class="nav-link {% if '/archive' in request.url.path %}active{% endif %}">Archive</a>
         <a href="/admin/settings" class="nav-link {% if '/settings' in request.url.path %}active{% endif %}">Settings</a>
         {% endif %}
         <button class="nav-link nav-donate-btn" onclick="document.getElementById('donateModal').classList.add('show')">Donate &#9829;</button>

codai/admin/templates/settings.html

@@ -69,6 +69,38 @@
     <span class="form-hint">Models will inherit this as default when configured</span>
   </div>
 </div>
+
+<!-- Archive -->
+<div class="card mb-0" style="margin-top:1rem">
+  <div class="card-title">Generation Archive</div>
+  <div class="form-row">
+    <label style="display:flex;align-items:center;gap:.5rem;cursor:pointer">
+      <input type="checkbox" id="s-arc-enabled">
+      <span style="font-size:13px;font-weight:500">Enable generation archive</span>
+    </label>
+    <span class="form-hint">When enabled, every generation (image, video, audio, text) is logged with its prompt, model, and output files.</span>
+  </div>
+  <div class="form-row">
+    <label class="form-label">Archive directory <span class="muted" id="s-arc-dir-hint"></span></label>
+    <input type="text" id="s-arc-dir" class="form-input" placeholder="(default)">
+    <span class="form-hint">Absolute path or relative to the config directory. Leave blank to use the default.</span>
+  </div>
+  <div class="form-row" style="margin:0">
+    <label class="form-label">Retention period</label>
+    <select id="s-arc-retention" class="form-input" style="max-width:200px">
+      <option value="1h">1 hour</option>
+      <option value="1d">1 day</option>
+      <option value="2d">2 days</option>
+      <option value="1w">1 week</option>
+      <option value="1m">1 month</option>
+      <option value="3m">3 months</option>
+      <option value="6m">6 months</option>
+      <option value="1y">1 year</option>
+      <option value="never">Never (keep forever)</option>
+    </select>
+    <span class="form-hint">Archived entries older than this are automatically deleted. Takes effect immediately on save.</span>
+  </div>
+</div>
 {% endblock %}
 
 {% block scripts %}
@@ -99,6 +131,18 @@ async function loadSettings(){
     document.getElementById('s-gguf-cache').value = d.models?.gguf_cache_dir ?? '';
     document.getElementById('s-offload-dir').value = d.offload?.directory ?? './offload';
     toggleHttps();
+    // Archive
+    const arc = d.archive || {};
+    document.getElementById('s-arc-enabled').checked = !!arc.enabled;
+    document.getElementById('s-arc-dir').value = arc.directory ?? '';
+    document.getElementById('s-arc-retention').value = arc.retention ?? 'never';
+    // Show effective default dir as placeholder hint
+    try {
+      const as = await fetch('/admin/api/archive-settings').then(r=>r.json());
+      const hint = document.getElementById('s-arc-dir-hint');
+      if (hint && as.default_directory) hint.textContent = `(default: ${as.default_directory})`;
+      document.getElementById('s-arc-dir').placeholder = as.default_directory || '(default)';
+    } catch(_){}
   }catch(e){ showAlert('error','Failed to load settings: '+e.message); }
 }
 
@@ -119,14 +163,19 @@ async function saveSettings(){
     },
     offload:{
       directory: document.getElementById('s-offload-dir').value.trim() || './offload',
-    }
+    },
+    archive:{
+      enabled: document.getElementById('s-arc-enabled').checked,
+      directory: document.getElementById('s-arc-dir').value.trim(),
+      retention: document.getElementById('s-arc-retention').value,
+    },
   };
   try{
     const r = await fetch('/admin/api/settings',{
       method:'POST', headers:{'Content-Type':'application/json'},
       body: JSON.stringify(data)
     });
-    if(r.ok) showAlert('info','Settings saved. Restart CoderAI to apply.');
+    if(r.ok) showAlert('info','Settings saved. Archive changes take effect immediately; restart CoderAI for other changes.');
     else{ const e=await r.json(); showAlert('error', e.detail||'Save failed'); }
   }catch(e){ showAlert('error','Error: '+e.message); }
 }

codai/api/app.py

@@ -66,8 +66,33 @@ def set_global_file_path_wrapper(path: str):
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     """Lifespan context manager for startup/shutdown."""
-    # Startup
+    import asyncio as _asyncio
+
+    async def _archive_cleanup_loop():
+        from codai.api.archive import archive_manager
+        while True:
+            await _asyncio.sleep(3600)
+            try:
+                archive_manager.run_cleanup()
+            except Exception:
+                pass
+
+    cleanup_task = _asyncio.create_task(_archive_cleanup_loop())
+    # Run initial cleanup on startup
+    try:
+        from codai.api.archive import archive_manager
+        archive_manager.run_cleanup()
+    except Exception:
+        pass
+
     yield
+
+    cleanup_task.cancel()
+    try:
+        await cleanup_task
+    except _asyncio.CancelledError:
+        pass
+
     # Shutdown
     multi_model_manager.cleanup()
     model_manager.cleanup()
@@ -101,6 +126,8 @@ from codai.api.voice_clone import router as voice_clone_router
 from codai.api.voice_convert import router as voice_convert_router
 from codai.api.faceswap import router as faceswap_router
 from codai.api.characters import router as characters_router
+from codai.api.spatial import router as spatial_router
+from codai.api.environments import router as environments_router
 from codai.admin.routes import router as admin_router
 
 # Import and add middleware
@@ -133,6 +160,8 @@ app.include_router(voice_clone_router)
 app.include_router(voice_convert_router)
 app.include_router(faceswap_router)
 app.include_router(characters_router)
+app.include_router(environments_router)
+app.include_router(spatial_router)
 app.include_router(admin_router)
 
 
@@ -166,4 +195,59 @@ async def get_file(filename: str):
         raise HTTPException(status_code=403, detail="Access denied")
     if not os.path.isfile(candidate):
         raise HTTPException(status_code=404, detail="File not found")
-    return FileResponse(candidate)
+    return FileResponse(candidate)
\ No newline at end of file
+
+
+_IMAGE_EXTS = {'.png', '.jpg', '.jpeg', '.webp', '.gif'}
+_VIDEO_EXTS = {'.mp4', '.webm', '.avi', '.mov'}
+_AUDIO_EXTS = {'.wav', '.mp3', '.ogg', '.flac', '.aac', '.m4a'}
+
+
+@app.get("/v1/archive")
+async def list_archive():
+    """List all generated files in the output directory."""
+    if not global_file_path or not os.path.isdir(global_file_path):
+        return {"files": []}
+    files = []
+    try:
+        names = os.listdir(global_file_path)
+    except OSError:
+        return {"files": []}
+    for fname in names:
+        fpath = os.path.join(global_file_path, fname)
+        if not os.path.isfile(fpath):
+            continue
+        ext = os.path.splitext(fname)[1].lower()
+        if ext in _IMAGE_EXTS:
+            ftype = 'image'
+        elif ext in _VIDEO_EXTS:
+            ftype = 'video'
+        elif ext in _AUDIO_EXTS:
+            ftype = 'audio'
+        else:
+            continue
+        stat = os.stat(fpath)
+        files.append({
+            'filename': fname,
+            'type': ftype,
+            'size': stat.st_size,
+            'created': stat.st_mtime,
+            'url': f'/v1/files/{fname}',
+        })
+    files.sort(key=lambda f: f['created'], reverse=True)
+    return {"files": files}
+
+
+@app.delete("/v1/archive/{filename}")
+async def delete_archive_file(filename: str):
+    """Delete a generated file from the output directory."""
+    if not global_file_path:
+        raise HTTPException(status_code=404, detail="File not found")
+    safe_base = os.path.realpath(global_file_path)
+    candidate = os.path.realpath(os.path.join(global_file_path, filename))
+    if not (candidate == safe_base or candidate.startswith(safe_base + os.sep)):
+        raise HTTPException(status_code=403, detail="Access denied")
+    if not os.path.isfile(candidate):
+        raise HTTPException(status_code=404, detail="File not found")
+    os.remove(candidate)
+    return {"deleted": filename}
\ No newline at end of file

codai/api/archive.py

+# CoderAI - OpenAI-compatible API server
+# Copyright (C) 2026 Stefy Lanza <stefy@nexlab.net>
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program. If not, see <https://www.gnu.org/licenses/>.
+
+"""Generation archive — persists all generated artifacts with metadata on disk."""
+
+import io
+import json
+import logging
+import os
+import shutil
+import threading
+import time
+import uuid
+from datetime import datetime
+from typing import Any, Dict, List, Optional, Tuple
+
+_log = logging.getLogger(__name__)
+
+RETENTION_OPTIONS = ["1h", "1d", "2d", "1w", "1m", "3m", "6m", "1y", "never"]
+
+_RETENTION_SECONDS: Dict[str, Optional[int]] = {
+    "1h":       3_600,
+    "1d":      86_400,
+    "2d":     172_800,
+    "1w":     604_800,
+    "1m":   2_592_000,   # 30 days
+    "3m":   7_776_000,   # 90 days
+    "6m":  15_552_000,   # 180 days
+    "1y":  31_536_000,   # 365 days
+    "never":      None,
+}
+
+
+def _safe_id(s: str) -> bool:
+    return bool(s) and "/" not in s and ".." not in s and "\\" not in s
+
+
+def _safe_filename(s: str) -> bool:
+    return bool(s) and "/" not in s and ".." not in s and "\\" not in s
+
+
+class ArchiveManager:
+    """Manages the on-disk generation archive."""
+
+    def __init__(self):
+        self.enabled: bool = True
+        self.directory: str = os.path.abspath("./archive")
+        self.retention: str = "never"
+        self._lock = threading.Lock()
+
+    def configure(self, enabled: bool, directory: str, retention: str):
+        self.enabled = enabled
+        self.directory = os.path.abspath(directory or "./archive")
+        self.retention = retention if retention in _RETENTION_SECONDS else "never"
+
+    def _gen_dir(self, gen_id: str) -> str:
+        return os.path.join(self.directory, gen_id)
+
+    # ------------------------------------------------------------------
+    # Save
+    # ------------------------------------------------------------------
+
+    def save_generation(
+        self,
+        gen_type: str,
+        endpoint: str,
+        model: Optional[str],
+        prompt: Optional[str],
+        params: Dict[str, Any],
+        artifacts: List[Tuple[bytes, str]],  # [(raw_bytes, file_ext), ...]
+    ) -> Optional[str]:
+        """Save a generation to archive.  Returns gen_id on success, None otherwise."""
+        if not self.enabled:
+            return None
+        try:
+            gen_id = f"{int(time.time())}_{uuid.uuid4().hex[:8]}"
+            gen_dir = self._gen_dir(gen_id)
+            os.makedirs(gen_dir, exist_ok=True)
+
+            saved_files: List[str] = []
+            preview_file: Optional[str] = None
+
+            for i, (data, ext) in enumerate(artifacts):
+                if not data:
+                    continue
+                fname = f"output_{i}.{ext}"
+                with open(os.path.join(gen_dir, fname), "wb") as fh:
+                    fh.write(data)
+                saved_files.append(fname)
+
+                if i == 0 and preview_file is None:
+                    try:
+                        preview_file = self._make_preview(data, ext, gen_dir)
+                    except Exception as e:
+                        _log.debug("Archive preview failed: %s", e)
+
+            meta: Dict[str, Any] = {
+                "id": gen_id,
+                "type": gen_type,
+                "endpoint": endpoint,
+                "model": model or "",
+                "prompt": prompt or "",
+                "parameters": params,
+                "files": saved_files,
+                "preview": preview_file or (saved_files[0] if saved_files else None),
+                "created_at": int(time.time()),
+                "created_at_iso": datetime.utcnow().isoformat() + "Z",
+            }
+            with open(os.path.join(gen_dir, "metadata.json"), "w") as fh:
+                json.dump(meta, fh, indent=2)
+
+            return gen_id
+        except Exception as e:
+            _log.warning("Archive save failed: %s", e)
+            return None
+
+    # ------------------------------------------------------------------
+    # Query
+    # ------------------------------------------------------------------
+
+    def list_entries(self, limit: int = 50, offset: int = 0) -> List[Dict]:
+        """Return metadata dicts, newest first."""
+        if not os.path.isdir(self.directory):
+            return []
+        entries: List[Dict] = []
+        try:
+            for name in os.listdir(self.directory):
+                meta_path = os.path.join(self.directory, name, "metadata.json")
+                if not os.path.isfile(meta_path):
+                    continue
+                try:
+                    with open(meta_path) as fh:
+                        entries.append(json.load(fh))
+                except Exception:
+                    pass
+        except Exception:
+            pass
+        entries.sort(key=lambda x: x.get("created_at", 0), reverse=True)
+        return entries[offset: offset + limit]
+
+    def count_entries(self) -> int:
+        if not os.path.isdir(self.directory):
+            return 0
+        count = 0
+        try:
+            for name in os.listdir(self.directory):
+                if os.path.isfile(os.path.join(self.directory, name, "metadata.json")):
+                    count += 1
+        except Exception:
+            pass
+        return count
+
+    def get_entry(self, gen_id: str) -> Optional[Dict]:
+        if not _safe_id(gen_id):
+            return None
+        meta_path = os.path.join(self._gen_dir(gen_id), "metadata.json")
+        if not os.path.isfile(meta_path):
+            return None
+        try:
+            with open(meta_path) as fh:
+                return json.load(fh)
+        except Exception:
+            return None
+
+    def delete_entry(self, gen_id: str) -> bool:
+        if not _safe_id(gen_id):
+            return False
+        gen_dir = self._gen_dir(gen_id)
+        if not os.path.isdir(gen_dir):
+            return False
+        try:
+            shutil.rmtree(gen_dir)
+            return True
+        except Exception:
+            return False
+
+    def get_file_path(self, gen_id: str, filename: str) -> Optional[str]:
+        if not _safe_id(gen_id) or not _safe_filename(filename):
+            return None
+        path = os.path.join(self._gen_dir(gen_id), filename)
+        return path if os.path.isfile(path) else None
+
+    # ------------------------------------------------------------------
+    # Cleanup
+    # ------------------------------------------------------------------
+
+    def run_cleanup(self):
+        """Remove entries older than the configured retention window."""
+        max_age = _RETENTION_SECONDS.get(self.retention)
+        if max_age is None:
+            return
+        if not os.path.isdir(self.directory):
+            return
+        cutoff = time.time() - max_age
+        deleted = 0
+        try:
+            for name in list(os.listdir(self.directory)):
+                meta_path = os.path.join(self.directory, name, "metadata.json")
+                try:
+                    with open(meta_path) as fh:
+                        meta = json.load(fh)
+                    if meta.get("created_at", time.time()) < cutoff:
+                        shutil.rmtree(os.path.join(self.directory, name), ignore_errors=True)
+                        deleted += 1
+                except Exception:
+                    pass
+        except Exception:
+            pass
+        if deleted:
+            _log.info("Archive cleanup: removed %d entry/entries", deleted)
+
+    # ------------------------------------------------------------------
+    # Preview generation
+    # ------------------------------------------------------------------
+
+    def _make_preview(self, data: bytes, ext: str, gen_dir: str) -> Optional[str]:
+        ext = ext.lower()
+        if ext in ("png", "jpg", "jpeg", "webp", "bmp", "gif"):
+            from PIL import Image
+            img = Image.open(io.BytesIO(data)).convert("RGB")
+            img.thumbnail((512, 512))
+            dst = os.path.join(gen_dir, "preview.webp")
+            img.save(dst, "WEBP", quality=75)
+            return "preview.webp"
+
+        if ext in ("mp4", "webm", "avi", "mov"):
+            import subprocess
+            import tempfile
+            with tempfile.NamedTemporaryFile(suffix=f".{ext}", delete=False) as tmp:
+                tmp.write(data)
+                tmp_path = tmp.name
+            frame_path = os.path.join(gen_dir, "_frame_tmp.jpg")
+            try:
+                r = subprocess.run(
+                    ["ffmpeg", "-y", "-i", tmp_path, "-vframes", "1", "-q:v", "3", frame_path],
+                    capture_output=True, timeout=15,
+                )
+                if r.returncode == 0 and os.path.exists(frame_path):
+                    from PIL import Image
+                    img = Image.open(frame_path).convert("RGB")
+                    img.thumbnail((512, 512))
+                    dst = os.path.join(gen_dir, "preview.webp")
+                    img.save(dst, "WEBP", quality=75)
+                    return "preview.webp"
+            except Exception:
+                pass
+            finally:
+                try:
+                    os.unlink(tmp_path)
+                except Exception:
+                    pass
+                try:
+                    os.unlink(frame_path)
+                except Exception:
+                    pass
+
+        # Audio or unrecognised — no visual preview
+        return None
+
+
+# Module-level singleton
+archive_manager = ArchiveManager()

codai/api/audio_gen.py

@@ -199,4 +199,24 @@ async def audio_generate(request: AudioGenerationRequest, http_request: Request
         raise HTTPException(status_code=500, detail=f"Audio generation failed: {e}")
 
     result = _save_audio_response(audio_bytes, ext, http_request)
+
+    try:
+        from codai.api.archive import archive_manager
+        asyncio.get_event_loop().create_task(asyncio.to_thread(
+            archive_manager.save_generation,
+            "audio", "/v1/audio/generate",
+            model_name,
+            request.prompt,
+            {
+                "duration": request.duration,
+                "top_k": request.top_k,
+                "top_p": request.top_p,
+                "temperature": request.temperature,
+                "cfg_coef": request.cfg_coef,
+            },
+            [(audio_bytes, ext)],
+        ))
+    except Exception:
+        pass
+
     return AudioGenerationResponse(created=int(time.time()), data=[result])
\ No newline at end of file

codai/api/ratelimit.py

@@ -34,9 +34,11 @@ from starlette.middleware.base import BaseHTTPMiddleware
 class BearerAuthMiddleware(BaseHTTPMiddleware):
     """Reject /v1/ API requests that lack a valid Bearer token or active web session."""
 
+    _EXEMPT_PATHS = {"/v1/images/progress"}
+
     async def dispatch(self, request: Request, call_next):
         path = request.url.path
-        if not path.startswith("/v1/"):
+        if not path.startswith("/v1/") or path in self._EXEMPT_PATHS:
             return await call_next(request)
 
         from codai.admin import routes as _admin_routes
@@ -112,14 +114,20 @@ class RateLimitMiddleware(BaseHTTPMiddleware):
                 return prefix
         return ""
 
+    # Lightweight polling endpoints that must never be rate-limited
+    _EXEMPT_PATHS = {"/v1/images/progress"}
+
     async def dispatch(self, request: Request, call_next):
         if not RATE_LIMITING_ENABLED:
             return await call_next(request)
 
         path = request.url.path
 
-        # Queue-size enforcement for authenticated API requests
-        if any(path.startswith(p) for p in _QUEUED_PREFIXES):
+        if path in self._EXEMPT_PATHS:
+            return await call_next(request)
+
+        # Queue-size enforcement for authenticated API requests (not for status polls)
+        if path not in self._EXEMPT_PATHS and any(path.startswith(p) for p in _QUEUED_PREFIXES):
             from codai.queue.manager import queue_manager
             if await queue_manager.is_full():
                 return JSONResponse(

codai/api/tts.py

@@ -18,10 +18,12 @@
 Text-to-speech endpoints for the codai API.
 """
 
+import asyncio
 import base64
 import os
+from typing import Optional
 
-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, HTTPException, Request
 from pydantic import BaseModel, ConfigDict
 
 # Import from codai modules
@@ -51,7 +53,8 @@ class TTSRequest(BaseModel):
     voice: str = "af_sarah"
     response_format: str = "mp3"
     speed: float = 1.0
-    
+    voice_profile: Optional[str] = None   # saved voice profile name (uses F5-TTS cloning)
+
     model_config = ConfigDict(extra="allow")
 
 
@@ -62,13 +65,39 @@ class TTSResponse(BaseModel):
 
 
 @router.post("/v1/audio/speech")
-async def create_speech(request: TTSRequest):
+async def create_speech(request: TTSRequest, http_request: Request = None):
     """
     Text-to-speech endpoint (OpenAI-compatible).
     
     Supports:
     - Kokoro TTS models (when --tts-model is specified)
     """
+    # If a voice profile is requested, delegate to voice cloning (F5-TTS)
+    if request.voice_profile:
+        from codai.api.voice_clone import _load_voice, _f5tts_clone
+        meta = _load_voice(request.voice_profile)
+        if not meta:
+            raise HTTPException(status_code=404,
+                detail=f"Voice profile '{request.voice_profile}' not found")
+        ref_audio_path = meta['audio_file']
+        ref_text = meta.get('transcript', '')
+        if not ref_text:
+            raise HTTPException(status_code=400,
+                detail="Voice profile has no transcript; update it with PATCH /v1/audio/voices/{name}")
+        try:
+            audio_bytes = await asyncio.get_event_loop().run_in_executor(
+                None, _f5tts_clone,
+                ref_audio_path, ref_text, request.input,
+                request.speed or 1.0, None,
+            )
+        except ImportError:
+            raise HTTPException(status_code=501,
+                detail="f5-tts not installed. Run: pip install f5-tts")
+        except Exception as e:
+            raise HTTPException(status_code=500, detail=f"Voice cloning failed: {e}")
+        audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
+        return {"audio": audio_base64}
+
     # Use the manager to resolve the model and manage VRAM
     model_info = multi_model_manager.request_model(
         requested_model=request.model,
@@ -119,10 +148,23 @@ async def create_speech(request: TTSRequest):
         speed = request.speed or 1.0
         
         audio_bytes = kokoro_model.generate(request.input, voice=voice, speed=speed)
-        
+
+        try:
+            from codai.api.archive import archive_manager
+            asyncio.get_event_loop().create_task(asyncio.to_thread(
+                archive_manager.save_generation,
+                "tts", "/v1/audio/speech",
+                model_name,
+                request.input,
+                {"voice": voice, "speed": speed, "response_format": request.response_format},
+                [(audio_bytes, request.response_format or "mp3")],
+            ))
+        except Exception:
+            pass
+
         # Convert to base64
         audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
-        
+
         return {
             "audio": audio_base64
         }

codai/api/voice_clone.py

 """
 Voice cloning endpoints.
 
-POST /v1/audio/clone          — synthesize speech in a cloned voice
-GET  /v1/audio/voices         — list saved voice profiles
-POST /v1/audio/voices         — save a named voice profile (ref audio + transcript)
-DELETE /v1/audio/voices/{name} — delete a voice profile
+POST   /v1/audio/clone              — synthesize speech in a cloned voice
+GET    /v1/audio/voices             — list saved voice profiles
+POST   /v1/audio/voices             — save a named voice profile (ref audio + transcript)
+POST   /v1/audio/voices/extract     — extract voice profile from audio or video file
+PATCH  /v1/audio/voices/{name}      — update description / replace reference audio
+DELETE /v1/audio/voices/{name}      — delete a voice profile
 """
 
 import asyncio
@@ -12,6 +14,7 @@ import base64
 import io
 import json
 import os
+import subprocess
 import tempfile
 import time
 from typing import Optional
@@ -103,6 +106,17 @@ def _decode_audio(data: str) -> tuple[bytes, str]:
     return base64.b64decode(data), '.wav'
 
 
+def _decode_b64_or_url(data: str) -> bytes:
+    if data.startswith('data:'):
+        _, b64 = data.split(',', 1)
+        return base64.b64decode(b64)
+    if data.startswith('http://') or data.startswith('https://'):
+        import urllib.request
+        with urllib.request.urlopen(data, timeout=60) as r:
+            return r.read()
+    return base64.b64decode(data)
+
+
 def _f5tts_clone(ref_audio_path: str, ref_text: str, gen_text: str,
                   speed: float = 1.0, seed: Optional[int] = None) -> bytes:
     """Run F5-TTS voice cloning, return WAV bytes."""
@@ -152,6 +166,26 @@ def _save_audio_response(audio_bytes: bytes, http_request: Request) -> dict:
     return {"b64_wav": base64.b64encode(audio_bytes).decode()}
 
 
+# ---------------------------------------------------------------------------
+# Pydantic models
+# ---------------------------------------------------------------------------
+
+class VoiceExtractRequest(BaseModel):
+    name: str
+    description: str = ''
+    audio: Optional[str] = None      # base64/URL audio file
+    video: Optional[str] = None      # base64/URL video (audio track extracted)
+    transcript: Optional[str] = ''   # optional; auto-transcribed if omitted
+    model_config = ConfigDict(extra="allow")
+
+
+class VoicePatchRequest(BaseModel):
+    description: Optional[str] = None
+    transcript: Optional[str] = None
+    audio: Optional[str] = None      # replace reference audio (base64)
+    model_config = ConfigDict(extra="allow")
+
+
 # ---------------------------------------------------------------------------
 # Voice profile management
 # ---------------------------------------------------------------------------
@@ -198,6 +232,122 @@ async def delete_voice(name: str):
     return {"deleted": True, "name": name}
 
 
+@router.patch("/v1/audio/voices/{name}")
+async def patch_voice(name: str, req: VoicePatchRequest):
+    """Update description, transcript, or reference audio of a saved voice profile."""
+    meta = _load_voice(name)
+    if not meta:
+        raise HTTPException(status_code=404, detail=f"Voice '{name}' not found")
+    vdir = _voice_path(name)
+
+    if req.description is not None:
+        meta['description'] = req.description
+    if req.transcript is not None:
+        meta['transcript'] = req.transcript
+    if req.audio:
+        audio_bytes, ext = _decode_audio(req.audio)
+        audio_file = os.path.join(vdir, f'ref{ext}')
+        # Remove old audio file(s)
+        for f in os.listdir(vdir):
+            if f.startswith('ref.') or f.startswith('ref'):
+                try:
+                    os.unlink(os.path.join(vdir, f))
+                except Exception:
+                    pass
+        with open(audio_file, 'wb') as f:
+            f.write(audio_bytes)
+        meta['audio_file'] = audio_file
+        meta['audio_ext'] = ext
+
+    with open(os.path.join(vdir, 'meta.json'), 'w') as f:
+        json.dump(meta, f)
+    return {"updated": True, "voice": meta}
+
+
+@router.get("/v1/audio/voices/{name}")
+async def get_voice(name: str):
+    """Get a single voice profile metadata."""
+    meta = _load_voice(name)
+    if not meta:
+        raise HTTPException(status_code=404, detail=f"Voice '{name}' not found")
+    return {"voice": meta}
+
+
+@router.post("/v1/audio/voices/extract")
+async def extract_voice(req: VoiceExtractRequest):
+    """
+    Extract a voice profile from a source audio or video file.
+
+    - audio: base64/URL audio file (wav, mp3, flac, …)
+    - video: base64/URL video file — audio track is extracted automatically
+    - transcript: optional; auto-transcribed with Whisper when omitted
+    """
+    if not req.name.replace('-', '').replace('_', '').isalnum():
+        raise HTTPException(status_code=400, detail="Voice name must be alphanumeric (hyphens/underscores allowed)")
+    if not req.audio and not req.video:
+        raise HTTPException(status_code=400, detail="Provide audio or video")
+
+    temps = []
+    try:
+        audio_bytes: Optional[bytes] = None
+        audio_ext = '.wav'
+
+        if req.video:
+            # Extract audio track from video with ffmpeg
+            raw = _decode_b64_or_url(req.video)
+            in_tmp = tempfile.NamedTemporaryFile(suffix='.mp4', delete=False)
+            in_tmp.write(raw)
+            in_tmp.close()
+            temps.append(in_tmp.name)
+            out_tmp = tempfile.NamedTemporaryFile(suffix='.wav', delete=False)
+            out_tmp.close()
+            temps.append(out_tmp.name)
+            r = subprocess.run(
+                ['ffmpeg', '-y', '-i', in_tmp.name, '-vn', '-acodec', 'pcm_s16le',
+                 '-ar', '22050', '-ac', '1', out_tmp.name],
+                capture_output=True, timeout=120,
+            )
+            if r.returncode != 0:
+                raise HTTPException(status_code=422,
+                    detail=f"Audio extraction from video failed: {r.stderr.decode(errors='replace')[:200]}")
+            with open(out_tmp.name, 'rb') as f:
+                audio_bytes = f.read()
+        else:
+            audio_bytes, audio_ext = _decode_audio(req.audio)
+
+        # Auto-transcribe if transcript not provided
+        transcript = req.transcript or ''
+        if not transcript:
+            try:
+                import whisper
+                in_tmp2 = tempfile.NamedTemporaryFile(suffix=audio_ext, delete=False)
+                in_tmp2.write(audio_bytes)
+                in_tmp2.close()
+                temps.append(in_tmp2.name)
+                model = whisper.load_model('base')
+                result = model.transcribe(in_tmp2.name)
+                transcript = result.get('text', '').strip()
+            except Exception:
+                pass
+
+        # Validate audio
+        try:
+            import soundfile as sf
+            sf.info(io.BytesIO(audio_bytes))
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=f"Invalid audio: {e}")
+
+        meta = _save_voice(req.name, audio_bytes, audio_ext, transcript, req.description)
+        return {"created": True, "voice": meta}
+
+    finally:
+        for t in temps:
+            try:
+                os.unlink(t)
+            except Exception:
+                pass
+
+
 # ---------------------------------------------------------------------------
 # Voice cloning TTS
 # ---------------------------------------------------------------------------

codai/config.py

@@ -98,6 +98,14 @@ class WhisperConfig:
     server_port: int = 8744
 
 
+@dataclass
+class ArchiveConfig:
+    """Generation archive configuration."""
+    enabled: bool = True
+    directory: str = ""        # empty = <config_dir>/archive; relative paths resolve from config_dir
+    retention: str = "never"   # one of: 1h 1d 2d 1w 1m 3m 6m 1y never
+
+
 @dataclass
 class Config:
     """Main configuration class."""
@@ -109,6 +117,7 @@ class Config:
     vulkan: VulkanConfig = field(default_factory=VulkanConfig)
     image: ImageConfig = field(default_factory=ImageConfig)
     whisper: WhisperConfig = field(default_factory=WhisperConfig)
+    archive: ArchiveConfig = field(default_factory=ArchiveConfig)
     system_prompt: Optional[str] = None
     tools_closer_prompt: bool = False
     grammar_guided: bool = False
@@ -244,6 +253,7 @@ class ConfigManager:
                 vulkan=VulkanConfig(**config_data.get("vulkan", {})),
                 image=ImageConfig(**config_data.get("image", {})),
                 whisper=WhisperConfig(**config_data.get("whisper", {})),
+                archive=ArchiveConfig(**config_data.get("archive", {})),
                 system_prompt=config_data.get("system_prompt"),
                 tools_closer_prompt=config_data.get("tools_closer_prompt", False),
                 grammar_guided=config_data.get("grammar_guided", False),
@@ -347,6 +357,11 @@ class ConfigManager:
                 "vae_tiling": self.config.image.vae_tiling,
                 "clip_on_cpu": self.config.image.clip_on_cpu
             },
+            "archive": {
+                "enabled": self.config.archive.enabled,
+                "directory": self.config.archive.directory,
+                "retention": self.config.archive.retention,
+            },
             "system_prompt": self.config.system_prompt,
             "tools_closer_prompt": self.config.tools_closer_prompt,
             "grammar_guided": self.config.grammar_guided,

codai/main.py

@@ -22,6 +22,7 @@ import os
 from codai.cli import parse_args
 from codai.config import ConfigManager
 from codai.admin.routes import init_session_manager, set_config_manager
+from codai.api.archive import archive_manager
 
 
 def main():
@@ -55,6 +56,18 @@ def main():
     if config.models.gguf_cache_dir:
         os.environ['CODERAI_CACHE_DIR'] = config.models.gguf_cache_dir
     
+    # Configure generation archive
+    _arc_dir = config.archive.directory
+    if not _arc_dir:
+        _arc_dir = os.path.join(config_dir, "archive")
+    elif not os.path.isabs(_arc_dir):
+        _arc_dir = os.path.join(config_dir, _arc_dir)
+    archive_manager.configure(
+        enabled=config.archive.enabled,
+        directory=_arc_dir,
+        retention=config.archive.retention,
+    )
+
     # Initialize admin session manager and expose config to admin routes
     from pathlib import Path
     init_session_manager(Path(config_dir))
@@ -364,6 +377,8 @@ def main():
                 multi_model_manager.model_backend_types[mid] = (
                     m.get("backend", "auto") if isinstance(m, dict) else "auto"
                 )
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Audio models
     audio_models = models_config.get("audio_models", [])
@@ -400,6 +415,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_image_model(mid, config=_model_cfg(m, "image"))
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Vision models
     vision_models = models_config.get("vision_models", [])
@@ -407,6 +424,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_vision_model(mid, config=_model_cfg(m, "vision"))
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # TTS models
     tts_models = models_config.get("tts_models", [])
@@ -414,6 +433,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_tts_model(mid, config=_model_cfg(m, "tts") if isinstance(m, dict) else {})
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Video generation models
     video_models = models_config.get("video_models", [])
@@ -421,6 +442,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_video_model(mid, config=_model_cfg(m, "video") if isinstance(m, dict) else {})
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Audio generation models (MusicGen, AudioLDM2, …)
     audio_gen_models = models_config.get("audio_gen_models", [])
@@ -428,6 +451,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_audio_gen_model(mid, config=_model_cfg(m, "audio_gen") if isinstance(m, dict) else {})
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Embedding models
     embedding_models = models_config.get("embedding_models", [])
@@ -435,6 +460,8 @@ def main():
         mid = _model_id(m)
         if mid:
             multi_model_manager.set_embedding_model(mid, config=_model_cfg(m, "embedding") if isinstance(m, dict) else {})
+            if isinstance(m, dict) and m.get("alias"):
+                multi_model_manager.set_model_alias(m["alias"], mid)
 
     # Register aliases
     aliases = models_config.get("aliases", {})
@@ -623,10 +650,20 @@ def main():
     if global_file_path:
         set_fs_file_path(global_file_path)
 
+    # Set spatial (3D) module global args
+    from codai.api.spatial import set_global_args as set_spatial_global_args, set_global_file_path as set_spatial_file_path
+    set_spatial_global_args(global_args)
+    if global_file_path:
+        set_spatial_file_path(global_file_path)
+
     # Set character profiles module global args
     from codai.api.characters import set_global_args as set_chars_global_args
     set_chars_global_args(global_args)
 
+    # Set environment profiles module global args
+    from codai.api.environments import set_global_args as set_envs_global_args
+    set_envs_global_args(global_args)
+
     # Set embeddings module global args
     from codai.api.embeddings import set_global_args as set_embed_global_args
     set_embed_global_args(global_args)

codai/models/capabilities.py

@@ -62,6 +62,12 @@ class ModelCapabilities:
     lip_sync: bool = False              # Wav2Lip, SadTalker
     video_dubbing: bool = False         # Translation + TTS + lip sync
 
+    # 3D generation and conversion
+    image_to_3d: bool = False           # 2D image → stereo / anaglyph / depth / mesh
+    video_to_3d: bool = False           # 2D video → stereo / anaglyph / depth video
+    model_3d_generation: bool = False   # text / image → 3D model (GLB)
+    model_3d_to_image: bool = False     # 3D model → rendered 2D image / video
+
     def to_list(self) -> List[str]:
         out = []
         for name, val in self.__dataclass_fields__.items():
@@ -85,6 +91,15 @@ def detect_model_capabilities(model_name: str) -> ModelCapabilities:
 
     n = model_name.lower()
 
+    # ── 3D generation ────────────────────────────────────────────────────────
+    if any(x in n for x in ['triposr', 'tsr', 'shap-e', 'shape-e', 'point-e', 'pointe',
+                              'zero123', 'wonder3d', 'instant3d', 'one-2-3-45',
+                              'mvdiffusion', 'syncdreamer', 'dreamgaussian']):
+        caps.model_3d_generation = True
+        caps.image_to_3d = True
+        caps.model_3d_to_image = True
+        return caps
+
     # ── Video generation ─────────────────────────────────────────────────────
     if any(x in n for x in ['cogvideox', 'cogvideo', 'ltx-video', 'ltxvideo',
                               'hunyuan-video', 'mochi-1', 'dynamicrafter',

codai/models/manager.py

@@ -1632,6 +1632,23 @@ class MultiModelManager:
                 'error': f"Model '{resolved_name}' is not available. Use one of: {', '.join(allowed_ids)}",
             }
         
+        # Step 1c: Resolve short-name to canonical registered name (e.g. "Foo" → "org/Foo")
+        _type_registry = {
+            "image": self.image_models,
+            "audio": self.audio_models,
+            "tts": [self.tts_model] if self.tts_model else [],
+            "vision": self.vision_models,
+            "video": self.video_models,
+            "audio_gen": self.audio_gen_models,
+            "embedding": self.embedding_models,
+        }
+        _candidates = _type_registry.get(model_type, []) if model_type else []
+        if resolved_name not in _candidates:
+            for _m in _candidates:
+                if "/" in _m and _m.split("/")[-1] == resolved_name:
+                    resolved_name = _m
+                    break
+
         # Step 2: Build the model key (prefixed with type)
         if model_type and model_type != "text":
             model_key = f"{model_type}:{resolved_name}"
@@ -1641,6 +1658,19 @@ class MultiModelManager:
         # Step 3: Check if already loaded in self.models
         existing_model = self.models.get(model_key)
 
+        # Fuzzy fallback: if not found by exact key, check for org-prefix
+        # variations (e.g. "image:Model" vs "image:org/Model").
+        if existing_model is None:
+            type_prefix = f"{model_type}:" if model_type and model_type != "text" else ""
+            short = resolved_name.split("/")[-1]
+            for k, v in self.models.items():
+                k_name = k[len(type_prefix):] if type_prefix and k.startswith(type_prefix) else k
+                if k_name == resolved_name or k_name.split("/")[-1] == short:
+                    model_key = k
+                    resolved_name = k_name
+                    existing_model = v
+                    break
+
         # =====================================================================
         # PER-MODEL LOAD MODE: Check per-model config first.
         # Per-model "load" = pre-loaded (treat as loadall for this model).

codai/pydantic/audiogenrequest.py

@@ -31,6 +31,8 @@ class AudioGenerationRequest(BaseModel):
     seed: Optional[int] = None
     # Reference audio for melody conditioning (MusicGen Melody)
     melody: Optional[str] = None           # base64/URL
+    # Voice profile for singing/speech conditioning
+    voice_profile: Optional[str] = None    # saved voice profile name
     # Output
     response_format: Optional[str] = "url"  # url | b64_wav | b64_mp3
     user: Optional[str] = None

codai/pydantic/imagerequest.py

@@ -34,6 +34,12 @@ class ImageGenerationRequest(BaseModel):
     seed: Optional[int] = None
     user: Optional[str] = None
     disable_safety_checker: Optional[bool] = False
+    negative_prompt: Optional[str] = None
+
+    # Character consistency
+    character_profiles: Optional[List[str]] = None      # saved profile names
+    character_references: Optional[List[str]] = None    # inline base64 images
+    character_strength: Optional[float] = 0.6           # IP-Adapter scale
 
     model_config = ConfigDict(extra="allow")
 

codai/pydantic/videorequest.py

@@ -20,6 +20,18 @@ from typing import Dict, List, Optional
 from pydantic import BaseModel, ConfigDict
 
 
+class CharacterDialogLine(BaseModel):
+    """One spoken line in a multi-character dialog sequence."""
+    character: Optional[str] = None    # character profile name (used for lip-sync face)
+    voice: Optional[str] = None        # voice profile name OR TTS voice id
+    text: str                          # dialog text to synthesise
+    start_time: Optional[float] = None # explicit offset in seconds (None = sequential)
+    lip_sync: Optional[bool] = True    # apply lip sync for this line
+    lang: Optional[str] = None         # language hint for TTS
+    speed: Optional[float] = 1.0
+    model_config = ConfigDict(extra="allow")
+
+
 class VideoGenerationRequest(BaseModel):
     model: str
     prompt: str = ""
@@ -78,6 +90,9 @@ class VideoGenerationRequest(BaseModel):
     lip_sync: Optional[bool] = False        # warp mouth to match audio
     lip_sync_method: Optional[str] = "wav2lip"  # wav2lip | sadtalker
 
+    # ── Multi-character dialog ────────────────────────────────────────────
+    dialogs: Optional[List[CharacterDialogLine]] = None
+
     # ── Subtitles ────────────────────────────────────────────────────────
     generate_subtitles: Optional[bool] = False
     burn_subtitles: Optional[bool] = False