Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity

add option to show intermediate latent space

Browse Source
This commit is contained in:
damian
2022-11-01 11:17:43 +01:00
committed by Lincoln Stein
parent be1393a41c
commit cdb107dcda
9 changed files with 751 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
ldm/invoke/generator/base.py
View File

@ -116,6 +116,29 @@ class Generator():
)
return Image.fromarray(x_sample.astype(np.uint8))
# write an approximate RGB image from latent samples for a single step to PNG
def sample_to_lowres_estimated_image(self,samples):
# adapted from code by @erucipe and @keturn here:
# https://discuss.huggingface.co/t/decoding-latents-to-rgb-without-upscaling/23204/7
# these numbers were determined empirically by @keturn
v1_4_latent_rgb_factors = torch.tensor([
# R G B
[ 0.298, 0.207, 0.208], # L1
[ 0.187, 0.286, 0.173], # L2
[-0.158, 0.189, 0.264], # L3
[-0.184, -0.271, -0.473], # L4
], dtype=samples.dtype, device=samples.device)
latent_image = samples[0].permute(1, 2, 0) @ v1_4_latent_rgb_factors
latents_ubyte = (((latent_image + 1) / 2)
.clamp(0, 1) # change scale from -1..1 to 0..1
.mul(0xFF) # to 0..255
.byte()).cpu()
return Image.fromarray(latents_ubyte.numpy())
def generate_initial_noise(self, seed, width, height):
initial_noise = None
if self.variation_amount > 0 or len(self.with_variations) > 0:

10
ldm/invoke/server.py
View File

@ -34,6 +34,7 @@ def build_opt(post_data, seed, gfpgan_model_exists):
setattr(opt, 'facetool_strength', float(post_data['facetool_strength']) if gfpgan_model_exists else 0)
setattr(opt, 'upscale', [int(post_data['upscale_level']), float(post_data['upscale_strength'])] if post_data['upscale_level'] != '' else None)
setattr(opt, 'progress_images', 'progress_images' in post_data)
setattr(opt, 'progress_latents', 'progress_latents' in post_data)
setattr(opt, 'seed', None if int(post_data['seed']) == -1 else int(post_data['seed']))
setattr(opt, 'threshold', float(post_data['threshold']))
setattr(opt, 'perlin', float(post_data['perlin']))
@ -227,8 +228,13 @@ class DreamServer(BaseHTTPRequestHandler):
# since rendering images is moderately expensive, only render every 5th image
# and don't bother with the last one, since it'll render anyway
nonlocal step_index
if opt.progress_images and step % 5 == 0 and step < opt.steps - 1:
image = self.model.sample_to_image(sample)
wants_progress_latents = opt.progress_latents
wants_progress_image = opt.progress_image and step % 5 == 0
if (wants_progress_image | wants_progress_latents) and step < opt.steps - 1:
image = self.model.sample_to_image(sample) if wants_progress_image \
else self.model.sample_to_lowres_estimated_image(sample)
step_index_padded = str(step_index).rjust(len(str(opt.steps)), '0')
name = f'{prefix}.{opt.seed}.{step_index_padded}.png'
metadata = f'{opt.prompt} -S{opt.seed} [intermediate]'