inference much slow!

[hotpot-killer]

when i using brushnet to outpainting task, it is much slow! why is it ?

the inference info is below:

Base model type: SD1.5
BrushNet image.shape = torch.Size([1, 1536, 1424, 3]) mask.shape = torch.Size([1, 1536, 1424])
Requested to load AutoencoderKL
Loading 1 new model
BrushNet CL: image_latents shape = torch.Size([1, 4, 192, 178]) interpolated_mask shape = torch.Size([1, 1, 192, 178])
You should add to positive prompt: "empty scene"
Requested to load SD1ClipModel
Loading 1 new model
Requested to load BaseModel
Loading 1 new model

0%| | 0/20 [00:00<?, ?it/s]BrushNet inference: do_classifier_free_guidance is True
BrushNet inference, step = 0: image batch = 1, got 2 latents, starting from 0
BrushNet inference: sample torch.Size([2, 4, 192, 178]) , CL torch.Size([2, 5, 192, 178]) dtype torch.float16

5%|▌ | 1/20 [01:36<30:26, 96.13s/it]
10%|█ | 2/20 [03:17<29:45, 99.19s/it]
15%|█▌ | 3/20 [04:58<28:22, 100.18s/it]
20%|██ | 4/20 [06:40<26:50, 100.64s/it]
25%|██▌ | 5/20 [08:21<25:13, 100.89s/it]
30%|███ | 6/20 [10:02<23:34, 101.06s/it]
35%|███▌ | 7/20 [11:44<21:55, 101.17s/it]
40%|████ | 8/20 [13:25<20:14, 101.23s/it]
45%|████▌ | 9/20 [15:06<18:33, 101.26s/it]
50%|█████ | 10/20 [16:48<16:52, 101.28s/it]
55%|█████▌ | 11/20 [18:29<15:11, 101.30s/it]
60%|██████ | 12/20 [20:10<13:30, 101.32s/it]
65%|██████▌ | 13/20 [21:52<11:49, 101.33s/it]
70%|███████ | 14/20 [23:33<10:08, 101.34s/it]
75%|███████▌ | 15/20 [25:15<08:26, 101.34s/it]
80%|████████ | 16/20 [26:56<06:45, 101.35s/it]
85%|████████▌ | 17/20 [28:37<05:04, 101.35s/it]
90%|█████████ | 18/20 [30:19<03:22, 101.36s/it]
95%|█████████▌| 19/20 [32:00<01:41, 101.35s/it]
100%|██████████| 20/20 [33:41<00:00, 101.36s/it]
100%|██████████| 20/20 [33:41<00:00, 101.09s/it]
Prompt executed in 2207.95 seconds

[hotpot-killer]

@nullquant

[nullquant]

This is probably due to big image size for you VRAM. How much VRAM do you have and what card do you use?