Very slow generation

[InspectaGadget]

Hi !

I download nodes and workflow, i re-set (if needed) the models and all is good.
Workflow starts with no error, and correctly arrives until the Ksampler. I used it on vanilla settings.
Then things are going veeeery slow, to the point i don't manage to be sure it's really working (i know animatediff is a long process)
I'm running on nvidia RTX 4070 SUPER with 32 GB of Ram, and comfui on run_nvidia_gpu.bat.
GPU usage is on 100% and VRAM on 99 %. At least Temp is keeping average.

Ksampler preview don't seems like to product coherent pictures and I'm a little embarrassed not to dare to complete the process but it seems very long for an uncertain result.
I'm starting from large pictures, but I feel like they are all the way resize to coherent SD1.5 size.

Isn't it some problems here ? Did i miss something or maybe it's an average coherent time of generation ?

Thanks by advance

got prompt
model_type EPS
Using xformers attention in VAE
Using xformers attention in VAE
Requested to load SD1ClipModel
Loading 1 new model
Requested to load CLIPVisionModelProjection
Loading 1 new model
[AnimateDiffEvo] - INFO - Loading motion module sd15_t2v_beta.ckpt via Gen2
ValueSchedule: 0 1.15
1 1.15
2 1.15
3 1.15
4 1.15
5 1.15
6 1.15
7 1.15
8 1.15
9 1.15
10 1.15
11 1.15
12 1.15
13 1.15
14 1.15
15 1.15
16 1.15
17 1.15
18 1.15
19 1.15
20 1.15
21 1.15
22 1.15
23 1.15
24 1.15
25 1.15
26 1.15
27 1.15
28 1.15
29 1.15
30 1.15
31 1.15
32 1.15
33 1.15
34 1.15
35 1.15
36 1.15
37 1.15
38 1.15
39 1.15
40 1.15
41 1.15
42 1.15
43 1.15
44 1.15
45 1.15
46 1.15
47 1.15
48 1.15
49 1.15
50 1.15
51 1.15
52 1.15
53 1.15
54 1.15
55 1.15
56 1.15
dtype: float64
Using xformers attention in VAE
Using xformers attention in VAE
Requested to load AutoencoderKL
Loading 1 new model
[AnimateDiffEvo] - INFO - Sliding context window activated - latents passed in (57) greater than context_length 16.
[AnimateDiffEvo] - INFO - Using motion module sd15_t2v_beta.ckpt:v2.
Requested to load BaseModel
Requested to load AnimateDiffModel
Requested to load SparseControlNet
Loading 3 new models
5%|████ | 1/20 [10:08<3:12:46, 608.78s/it]

[lavibeats]

yeah only 16 frame around 35 minutes ! (rtx3090 )

[lealcc]

It is incredibly slow at mid-vram configs and just slow at 24gbs

[tampadesignr]

curious if new nodes and updates to comfyui and models to swapout nodes models etc.. to new stuff to speedup
lighttricks? might have some helpful stuff
dunno... but yes its a stand still not moving workflow right now