diff options
| author | catboxanon <122327233+catboxanon@users.noreply.github.com> | 2023-08-13 08:16:48 +0000 |
|---|---|---|
| committer | catboxanon <122327233+catboxanon@users.noreply.github.com> | 2023-08-13 08:16:48 +0000 |
| commit | 822597db49218de17e105e62075096284dfcfd41 (patch) | |
| tree | 9d83c9389d99f262d897ce28ebcac6a0e3089415 /modules/sd_samplers_common.py | |
| parent | da80d649fd6a6083be02aca5695367bd25abf0d5 (diff) | |
| download | stable-diffusion-webui-gfx803-822597db49218de17e105e62075096284dfcfd41.tar.gz stable-diffusion-webui-gfx803-822597db49218de17e105e62075096284dfcfd41.tar.bz2 stable-diffusion-webui-gfx803-822597db49218de17e105e62075096284dfcfd41.zip | |
Encode batches separately
Significantly reduces VRAM.
This makes encoding more inline with how decoding currently functions.
Diffstat (limited to 'modules/sd_samplers_common.py')
| -rw-r--r-- | modules/sd_samplers_common.py | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/modules/sd_samplers_common.py b/modules/sd_samplers_common.py index 09d1e11e..f9d034ca 100644 --- a/modules/sd_samplers_common.py +++ b/modules/sd_samplers_common.py @@ -92,7 +92,15 @@ def images_tensor_to_samples(image, approximation=None, model=None): model = shared.sd_model
image = image.to(shared.device, dtype=devices.dtype_vae)
image = image * 2 - 1
- x_latent = model.get_first_stage_encoding(model.encode_first_stage(image))
+ if len(image) > 1:
+ x_latent = torch.stack([
+ model.get_first_stage_encoding(
+ model.encode_first_stage(torch.unsqueeze(img, 0))
+ )[0]
+ for img in image
+ ])
+ else:
+ x_latent = model.get_first_stage_encoding(model.encode_first_stage(image))
return x_latent
|