FunUI Diffusion Model Loader

Load a diffusion model, choosing precision and attention backend.

By digital-garbage·Created about a year ago·Updated about 12 hours ago· 20

model
status

◄model_name▾►

◄weight_dtypedefault►

◄compute_dtypedefault►

◄attentiondefault►

◄fp16_accumulationfalse►

◄slafalse►

◄sla_sparsity0.90►

◄sla_block_size64►

◄sla_protect_audiotrue►

◄sla_min_seq_len8192►

◄sla_dense_last_steps0►

◄int8_convrotfalse►

◄sla_enginecomfy_kitchen►

◄sla_dense_steps1►

◄sla_referencesoff►

◄sla_tailfalse►

◄sla_stabilize_motionfalse►

◄sla_methodsla►

◄sla_tau1.30►

CategoryFunUI/Loaders

Inputs (19)

Name	Type	Default	Description
model_name	COMBO		0 options:
weight_dtype	COMBO	default	How the weights are stored. 'default' keeps whatever the file already is.
compute_dtype	COMBO	default	What the maths runs in, independently of storage.
attention	COMBO	default	Attention backend. Only what this machine can run is listed; 'default' keeps ComfyUI's launch choice.
fp16_accumulationopt	BOOLEAN	false	Faster fp16 matmuls where the torch build supports it.
slaopt	BOOLEAN	false	Block-sparse attention for MiniMax H3 -- the inference path lightx2v's SLA turbo LoRA was distilled against, which is why that LoRA gives no speedup on its own. Roughly 3.7x the attention throughput at 768p/15s. Runs ALONGSIDE the attention backend above rather than replacing it: SLA takes H3's long packed self-attention, the chosen backend takes the text refiner, masked calls and any trailing dense steps. Skipped, with the reason, on anything that is not MiniMax H3 or on a machine without Triton.
sla_sparsityopt	FLOAT	0.900–0.95	Fraction of key blocks skipped while SLA is on (of video cubes for method vsa). 0.90 is validated; 0.85 is lightx2v's own value and ~15% slower. Break-even is around 0.60 -- below that the kernel is SLOWER than dense, so a low value is a loss, not a safe fallback.
sla_block_sizeopt	COMBO	64	How many sequence tokens share one key selection. H3 packs audio at 80 rows per second, so a 128-row block forces 1.6s of speech down one attention pattern while the same rows are 3% of a video frame. Use 128 only when the audio does not matter; 32 routes finer still for about the same time. Triton engine only.
sla_protect_audioopt	BOOLEAN	true	Always attend the prompt's words and every audio stream, whatever top-k picks. Audio is ~1% of the packed sequence, so plain top-k regularly drops all of it and the soundtrack degrades while the video still looks fine. Reference images stay sparse: see sla_references.
sla_min_seq_lenopt	INT	81920–1000000	Sequences shorter than this stay dense. Guards the short text refiner, which must never be sparsified, and low-resolution runs where block selection costs more than it saves.
sla_dense_last_stepsopt	INT	00–8	Run the last N sampling steps at full attention. 0 matches lightx2v; 1 was tested and did not help, for +20% time.
int8_convrotopt	BOOLEAN	false	MiniMax H3 from an ordinary (bf16/fp16) file: store the attention and MLP weights as int8 (rotated first, so no lone outlier spoils the rounding), the way Comfy-Org's own int8_convrot files are. About half the memory and int8 matmuls on GPUs that have them; slightly different pictures. Does nothing to a file that is already quantized or is not H3 (the status says so), and is skipped when a weight dtype is forced above. The memory saving always holds; the speed-up is lost on layers a LoRA is applied to and when compute dtype is forced.
sla_engineopt	COMBO	comfy_kitchen	What runs the sparse calls. comfy_kitchen: ComfyUI's compiled int8 sol_attn (needs comfy-kitchen 0.2.32+; the Triton kernel runs instead where it is missing, and the status says so). triton: FunUI's own kernel, the only one with block size and stabilize motion.
sla_dense_stepsopt	STRING	1	Steps at full attention, counted from 1: '1' is the first step, '1-2' the first two, '1,6' the first and sixth; blank or '0', none. The first steps set the layout and how closely the prompt is followed, so the first is exact by default. Adds to dense last steps. The log names the steps that ran this way.
sla_referencesopt	COMBO	off	Reference images, conditioning frames and the prompt's vision tokens. off: sparse like the video. light: each keeps its best 15% for every query block. heavy: all of them exact, slow with big references. On comfy_kitchen, light picks one 15% for all query blocks.
sla_tailopt	BOOLEAN	false	comfy_kitchen engine: the skipped blocks still count, as one averaged term per query block, instead of not at all. Off matches what the SLA turbo LoRA was trained against; untested on H3.
sla_stabilize_motionopt	BOOLEAN	false	Triton engine: each layer leans toward the blocks it picked last step, so near ties stop flipping and showing as a faint double exposure on fast motion. A fix for that one symptom; a little more memory.
sla_methodopt	COMBO	sla	How the blocks worth attending are chosen. sla: a fixed share (sla_sparsity), what SLA turbo LoRAs were trained against. sol-attn: a per-block threshold (sla_tau), no training needed; comfy_kitchen engine. vsa: video cubes, only for models trained for it (the status says when the model lacks VSA's layers); run by ComfyUI's own Model Sparse Attention, keeping 1 - sla_sparsity of the cubes.
sla_tauopt	FLOAT	1.300–4	sol-attn only: higher skips more. About 1.0 keeps ~16% of key blocks exact, 1.5 ~7%, 2.0 ~3%.

Outputs (2)

Name	Type	Description
model	MODEL	—
status	STRING	—