Commit Graph
35 Commits
Author SHA1 Message Date
TowstyandCursor 95873744cd Size Qwen 2.1 Generate from the bench aspect table.
Wire EmptyLatentImage to the official ~1 MP grid so 16:9 lands at 1536×864, keep PE wh_ratio advisory, and leave Edit on the encode latent from image_1.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-26 10:54:18 -05:00
TowstyandCursor e0292c8d2a Add Qwen 2.1 PE-only Comfy graphs for prompt enhance.
Keep PE CLIP separate from DiT so the 16 GB card can run rewrite then generate without loading both at once.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-26 10:25:44 -05:00
TowstyandCursor 8609ba24a4 Add Qwen 2.1 Image Edit Comfy graph (GGUF + TextEncode + int8 cache).
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-20 18:10:50 -05:00
TowstyandCursor 608f7d8d47 Fix Qwen 2.1 GGUF load: promote Q8 norms to F32 and wire TextEncode latent.
abenzerps Q8_0 ships 1D RMSNorms as packed Q8 (136 vs 128), which breaks
Comfy rms_rope; tagger now dequantizes small tensors and the graph uses
TextEncodeQwenImage21's 64-ch latent plus AuraFlow shift.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-20 16:58:00 -05:00
TowstyandCursor 08d507c491 Add studio2_qwen21_t2i Comfy graph for GGUF Qwen Image 2.1 T2I.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-20 16:57:04 -05:00
Towsty bd82f5bd5f Add native H3 identity graph and Studio 2 prompt contracts 2026-09-10 20:13:53 -05:00
Towsty 7665285a0d Preserve extension quality with original segments and reliable refinement 2026-09-08 06:25:40 -05:00
Towsty c2b6032e49 Replace Comfy YuE with standalone YuEGP profile 1 2026-09-07 20:24:47 -05:00
TowstyandCursor f41ac7a826 Add extension-frame refine and lossless anchor options.
Prefer clip anchor PNGs for handoffs, optionally run Bilateral+Flux cleanup before I2V, and save decoded last-frame anchors so sequential extensions accumulate less compression damage.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-04 07:55:46 -05:00
TowstyandCursor 17d4d4ffdc Tighten YuE lyrics validation and Comfy music workflow wiring.
Require section tags before queueing YuE and keep the ACE/YuE path changes that were already in progress.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-03 20:32:41 -05:00
TowstyandCursor de433c40d3 Lock YuE to a 16GB Comfy-safe recipe instead of chasing single knobs.
Keep mmgp profile 2 (no quanto), node-legal token floor, two Stage-A segments, and document why profiles 3-5 blow up under InferenceMode.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-03 17:46:54 -05:00
TowstyandCursor 98253b124d Keep YuE max_new_tokens at the node minimum of 2944.
Comfy rejects anything lower, so the 16GB VRAM tweak cannot go under that floor.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-03 17:35:46 -05:00
TowstyandCursor 3081b8fa36 Tighten YuE for 16GB VRAM and stop the false Waiting status.
Cap Stage-A segments and use a heavier mmgp profile so long lyrics stop OOMing mid-run, and show that Comfy is actually working.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-03 17:19:19 -05:00
TowstyandCursor 78b158471b Add ACE-Step 1.5 and YuE on Music, and put Video/Image/Music cards back.
Recover finished Comfy audio when queueing hangs so a completed track is not stuck on Queueing.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-03 07:42:54 -05:00
TowstyandCursor 5555155884 Add ACE-Step music generation to the studio and library.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-02 23:58:18 -05:00
TowstyandCursor 0fa64ea765 Keep last-frame continuity when character stills are attached.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-30 08:21:57 -05:00
TowstyandCursor 5fe91e6bae Show Flux and Krea as Image v2 engine cards, and run Krea on edit, compose, and refine.
Krea image-to-image uses encode plus denoise graphs, not Klein reference latents, and still fails closed if the Krea models are missing.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-29 07:29:31 -05:00
TowstyandCursor f6663444f5 Add Krea as a second Image v2 Generate engine beside Flux.
Generate can queue krea_v2_generate.json without sharing Klein encoders or falling back if Krea assets are missing.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-29 07:05:01 -05:00
TowstyandCursor b35f9af10d Add Image v2 Generate for Klein text-to-image.
New sibling graph and mode so T2I does not borrow Edit, Compose, or Refine. Stills are ignored; size is width x height only.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-28 22:56:33 -05:00
TowstyandCursor a2895b8054 Add Image v2 Refine for on-device regional edits.
Mask painter and denoise strength sit beside Edit and Compose. The canned hand/chest prompt helper is gone so that text stays yours.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-28 22:03:20 -05:00
TowstyandCursor 21fe2412f9 Add Image v2 as a sibling of Image edit, with separate Klein graphs.
Keep the v1 Image tab on its dual-branch template. The host agent follows Desktop's live port so Coolify stays on :8198.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-28 21:22:37 -05:00
TowstyandCursor 099b62edd7 Add a Scale to 1 MP toggle on image edit before Klein.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-28 16:36:15 -05:00
TowstyandCursor f51c3772e5 Load only the LoRAs the user picked on Klein; do not require snofs.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-28 16:21:32 -05:00
TowstyandCursor e3cf2f4b53 Load the real Gemma Q4_K_XL GGUF for PinkCherry DualCLIP.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 19:55:46 -05:00
TowstyandCursor 67519046a0 Point LTX DualCLIP at the Gemma GGUF that is actually on Beast.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 19:46:17 -05:00
TowstyandCursor 40cd17f35c Add MiniMax and PinkCherry LTX as pickable video engines, with still or text start.
Shot 1 can run text-to-video; later shots still extend from the last frame. Recommend output is also normalized into shot-script cards.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 19:38:53 -05:00
TowstyandCursor 2bebeea526 Send Qwen VL sampling args so Recommend does not fail on missing top_k/top_p/min_p/repeat_penalty.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 17:17:51 -05:00
TowstyandCursor df5ac24622 Add sidecar Qwen VL prompt recommend, fix queued shot bursts, and leave unused MiniMax identity slots empty.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 17:04:43 -05:00
TowstyandCursor 13ecec949c Run image edit on Beast with Flux.2 Klein and stop routing jobs to Sidecar.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-27 09:06:50 -05:00
TowstyandCursor 054c07ae03 Fix V2 identity queue validation by filling unused ref slots.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-26 23:18:45 -05:00
TowstyandCursor e334cb7a7a Move library collections as a group and map v2 identity refs.
Collapsed collection actions now move, rename, delete, and rerun every part so chains stay together. Identity mode follows the v2u3 Ref2VA picture wiring.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-26 22:56:37 -05:00
TowstyandCursor c55a48a57e Ship MiniMax v2 identity refs and Qwen sidecar edits.
Keep v1 as the default video graph, relay image edits to the Qwen Lightning box, and keep /api/health up if one Comfy host is unset.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-26 14:29:15 -05:00
TowstyandCursor f5829995c8 Start ComfyUI automatically before queueing MiniMax jobs.
Health-check /system_stats, wait if the desktop is booting or busy, and start the host instance when it is fully offline.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-25 19:39:28 -05:00
TowstyandCursor 30669c7a27 Swap in the custom MiniMax H3 graph and bind prompts, seed, size, steps, and LoRA to its node IDs.
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-08-25 18:02:25 -05:00
towsty 4c5c65db98 Add Nuxt MiniMax H3 relay studio for local ComfyUI.
Headless image-to-video UI queues jobs on the desktop ComfyUI instance with live SSE progress and Authentik SSO.
2026-08-25 17:43:29 -05:00