Repository Issue Activity (beta)

#1540 ComfyUI backend now requires PyTorch ≥ 2.7 with comfy-kitchen, instant crash on legacy hardware

#1537 Error while deserializing header: incomplete metadata, file not fully covered - Minimax H3 HL2VA

#1535 MiniMax H3 T2V: intermittent IndexError: list index out of range in SwarmKSampler (comfy/ldm/minimax/model.py:564, audio_src = x[1])

#876 Dynamic LoRa weight control, Start and stop LoRa at step count

#1534 Renorm CFG is being also applied to seedvr model

#1499 Video End Image Not Scaled Appropriately in Video Models

#1533 Minimax Ref2VA with video input errors out

#1531 Doesn't install dependencies properly on CachyOS with AMD ROCm

#1441 Krea 2 Prompt Weighting

#1528 Legacy prompt parsing strips `\(` escapes that ComfyUI's tokenizer still needs

#1527 LLM Spam

#1473 Video Reference FPS Fix Button

#1509 Feature Request: add gear menu option to pause videos in the batch view

#976 Allow loading same model twice in memory for refiner, avoid speed penalty when using <base> and <refiner>

#1520 Add minimax h3 default VAE override in User->User settings

#1522 Add "Use as init" to the expanded image view

#1521 Feature Request: Add image seed override to Refine/Upscale tab

#1519 Experimental editor copy-paste glitch

#1517 Z-Image default FP8 model fails on macOS/MPS with Undefined type Float8_e4m3fn

#1494 LTX 2.5 error: ComfyUI execution error: ERROR: VAE is invalid: None

#1510 Feature: Single-frame latent for H3

#1513 Spam

#1514 Spam

#1511 Show megapixel scale in places resolution info is displayed

#1512 Show resulting output resolution after up/downscaling