technicalkiwi
/
latentblending
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

hard fixed params for sdxl turbo

This commit is contained in:
Johannes Stelzer 2024-01-08 20:56:24 +01:00
parent df02d15562
commit 87aba9a8dc
1 changed files with 49 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

93
latent_blending.py
View File

@ -89,9 +89,9 @@ class LatentBlending():
self.list_nmb_branches = None self.list_nmb_branches = None
# Mixing parameters # Mixing parameters
self.branch1_crossfeed_power = 0.3 self.branch1_crossfeed_power = 0.0
self.branch1_crossfeed_range = 0.3 self.branch1_crossfeed_range = 0.0
self.branch1_crossfeed_decay = 0.99 self.branch1_crossfeed_decay = 0.0
self.parental_crossfeed_power = 0.3 self.parental_crossfeed_power = 0.3
self.parental_crossfeed_range = 0.6 self.parental_crossfeed_range = 0.6
@ -101,7 +101,6 @@ class LatentBlending():
self.multi_transition_img_first = None self.multi_transition_img_first = None
self.multi_transition_img_last = None self.multi_transition_img_last = None
self.dt_per_diff = 0 self.dt_per_diff = 0
self.spatial_mask = None
self.lpips = lpips.LPIPS(net='alex').cuda(self.device) self.lpips = lpips.LPIPS(net='alex').cuda(self.device)
self.set_prompt1("") self.set_prompt1("")
@ -215,6 +214,8 @@ class LatentBlending():
recycle_img1: Optional[bool] = False, recycle_img1: Optional[bool] = False,
recycle_img2: Optional[bool] = False, recycle_img2: Optional[bool] = False,
num_inference_steps: Optional[int] = 30, num_inference_steps: Optional[int] = 30,
list_idx_injection: Optional[int] = None,
list_nmb_stems: Optional[int] = None,
depth_strength: Optional[float] = 0.3, depth_strength: Optional[float] = 0.3,
t_compute_max_allowed: Optional[float] = None, t_compute_max_allowed: Optional[float] = None,
nmb_max_branches: Optional[int] = None, nmb_max_branches: Optional[int] = None,
@ -249,6 +250,7 @@ class LatentBlending():
assert self.text_embedding1 is not None, 'Set the first text embedding with .set_prompt1(...) before' assert self.text_embedding1 is not None, 'Set the first text embedding with .set_prompt1(...) before'
assert self.text_embedding2 is not None, 'Set the second text embedding with .set_prompt2(...) before' assert self.text_embedding2 is not None, 'Set the second text embedding with .set_prompt2(...) before'
# Random seeds # Random seeds
if fixed_seeds is not None: if fixed_seeds is not None:
if fixed_seeds == 'randomize': if fixed_seeds == 'randomize':
@ -260,6 +262,8 @@ class LatentBlending():
self.seed2 = fixed_seeds[1] self.seed2 = fixed_seeds[1]
# Ensure correct num_inference_steps in holder # Ensure correct num_inference_steps in holder
if 'turbo' in self.dh.pipe._name_or_path:
num_inference_steps = 4 #ideal results
self.num_inference_steps = num_inference_steps self.num_inference_steps = num_inference_steps
self.dh.set_num_inference_steps(num_inference_steps) self.dh.set_num_inference_steps(num_inference_steps)
@ -281,11 +285,18 @@ class LatentBlending():
self.tree_final_imgs = [self.dh.latent2image((self.tree_latents[0][-1])), self.dh.latent2image((self.tree_latents[-1][-1]))] self.tree_final_imgs = [self.dh.latent2image((self.tree_latents[0][-1])), self.dh.latent2image((self.tree_latents[-1][-1]))]
self.tree_idx_injection = [0, 0] self.tree_idx_injection = [0, 0]
# Hard-fix. Apply spatial mask only for list_latents2 but not for transition. WIP...
self.spatial_mask = None
# Set up branching scheme (dependent on provided compute time) # Set up branching scheme (dependent on provided compute time)
list_idx_injection, list_nmb_stems = self.get_time_based_branching(depth_strength, t_compute_max_allowed, nmb_max_branches) if 'turbo' in self.dh.pipe._name_or_path:
self.guidance_scale = 0.0
self.parental_crossfeed_power = 1.0
self.parental_crossfeed_power_decay = 1.0
self.parental_crossfeed_range = 1.0
list_idx_injection = [2]
list_nmb_stems = [10]
else:
list_idx_injection, list_nmb_stems = self.get_time_based_branching(depth_strength, t_compute_max_allowed, nmb_max_branches)
# Run iteratively, starting with the longest trajectory. # Run iteratively, starting with the longest trajectory.
# Always inserting new branches where they are needed most according to image similarity # Always inserting new branches where they are needed most according to image similarity
@ -298,7 +309,7 @@ class LatentBlending():
self.set_guidance_mid_dampening(fract_mixing) self.set_guidance_mid_dampening(fract_mixing)
list_latents = self.compute_latents_mix(fract_mixing, b_parent1, b_parent2, idx_injection) list_latents = self.compute_latents_mix(fract_mixing, b_parent1, b_parent2, idx_injection)
self.insert_into_tree(fract_mixing, idx_injection, list_latents) self.insert_into_tree(fract_mixing, idx_injection, list_latents)
# print(f"fract_mixing: {fract_mixing} idx_injection {idx_injection}") # print(f"fract_mixing: {fract_mixing} idx_injection {idx_injection} bp1 {b_parent1} bp2 {b_parent2}")
return self.tree_final_imgs return self.tree_final_imgs
@ -417,8 +428,10 @@ class LatentBlending():
results. Use this if you want to have controllable results independent results. Use this if you want to have controllable results independent
of your computer. of your computer.
""" """
idx_injection_base = int(round(self.num_inference_steps * depth_strength)) idx_injection_base = int(np.floor(self.num_inference_steps * depth_strength))
list_idx_injection = np.arange(idx_injection_base, self.num_inference_steps - 1, 3)
steps = int(np.ceil(self.num_inference_steps/10))
list_idx_injection = np.arange(idx_injection_base, self.num_inference_steps, steps)
list_nmb_stems = np.ones(len(list_idx_injection), dtype=np.int32) list_nmb_stems = np.ones(len(list_idx_injection), dtype=np.int32)
t_compute = 0 t_compute = 0
@ -440,7 +453,7 @@ class LatentBlending():
t_compute += 2 * self.num_inference_steps * self.dt_per_diff # outer branches t_compute += 2 * self.num_inference_steps * self.dt_per_diff # outer branches
increase_done = False increase_done = False
for s_idx in range(len(list_nmb_stems) - 1): for s_idx in range(len(list_nmb_stems) - 1):
if list_nmb_stems[s_idx + 1] / list_nmb_stems[s_idx] >= 2: if list_nmb_stems[s_idx + 1] / list_nmb_stems[s_idx] >= 1:
list_nmb_stems[s_idx] += 1 list_nmb_stems[s_idx] += 1
increase_done = True increase_done = True
break break
@ -471,15 +484,14 @@ class LatentBlending():
the index in terms of diffusion steps, where the next insertion will start. the index in terms of diffusion steps, where the next insertion will start.
""" """
# get_lpips_similarity # get_lpips_similarity
similarities = [] similarities = self.get_tree_similarities()
for i in range(len(self.tree_final_imgs) - 1):
similarities.append(self.get_lpips_similarity(self.tree_final_imgs[i], self.tree_final_imgs[i + 1]))
b_closest1 = np.argmax(similarities) b_closest1 = np.argmax(similarities)
b_closest2 = b_closest1 + 1 b_closest2 = b_closest1 + 1
fract_closest1 = self.tree_fracts[b_closest1] fract_closest1 = self.tree_fracts[b_closest1]
fract_closest2 = self.tree_fracts[b_closest2] fract_closest2 = self.tree_fracts[b_closest2]
fract_mixing = (fract_closest1 + fract_closest2) / 2
# Ensure that the parents are indeed older! # Ensure that the parents are indeed older
b_parent1 = b_closest1 b_parent1 = b_closest1
while True: while True:
if self.tree_idx_injection[b_parent1] < idx_injection: if self.tree_idx_injection[b_parent1] < idx_injection:
@ -492,7 +504,6 @@ class LatentBlending():
break break
else: else:
b_parent2 += 1 b_parent2 += 1
fract_mixing = (fract_closest1 + fract_closest2) / 2
return fract_mixing, b_parent1, b_parent2 return fract_mixing, b_parent1, b_parent2
def insert_into_tree(self, fract_mixing, idx_injection, list_latents): def insert_into_tree(self, fract_mixing, idx_injection, list_latents):
@ -507,10 +518,11 @@ class LatentBlending():
list of the latents to be inserted list of the latents to be inserted
""" """
b_parent1, b_parent2 = self.get_closest_idx(fract_mixing) b_parent1, b_parent2 = self.get_closest_idx(fract_mixing)
self.tree_latents.insert(b_parent1 + 1, list_latents) idx_tree = b_parent1 + 1
self.tree_final_imgs.insert(b_parent1 + 1, self.dh.latent2image(list_latents[-1])) self.tree_latents.insert(idx_tree, list_latents)
self.tree_fracts.insert(b_parent1 + 1, fract_mixing) self.tree_final_imgs.insert(idx_tree, self.dh.latent2image(list_latents[-1]))
self.tree_idx_injection.insert(b_parent1 + 1, idx_injection) self.tree_fracts.insert(idx_tree, fract_mixing)
self.tree_idx_injection.insert(idx_tree, idx_injection)
def get_noise(self, seed): def get_noise(self, seed):
r""" r"""
@ -807,6 +819,12 @@ class LatentBlending():
lploss = float(lploss[0][0][0][0]) lploss = float(lploss[0][0][0][0])
return lploss return lploss
def get_tree_similarities(self):
similarities = []
for i in range(len(self.tree_final_imgs) - 1):
similarities.append(self.get_lpips_similarity(self.tree_final_imgs[i], self.tree_final_imgs[i + 1]))
return similarities
# Auxiliary functions # Auxiliary functions
def get_closest_idx( def get_closest_idx(
self, self,
@ -832,7 +850,7 @@ class LatentBlending():
return b_parent1, b_parent2 return b_parent1, b_parent2
#%%
if __name__ == "__main__": if __name__ == "__main__":
# %% First let us spawn a stable diffusion holder. Uncomment your version of choice. # %% First let us spawn a stable diffusion holder. Uncomment your version of choice.
@ -841,22 +859,16 @@ if __name__ == "__main__":
from diffusers import AutoencoderTiny from diffusers import AutoencoderTiny
pipe = DiffusionPipeline.from_pretrained("stabilityai/sdxl-turbo", torch_dtype=torch.float16, variant="fp16") pipe = DiffusionPipeline.from_pretrained("stabilityai/sdxl-turbo", torch_dtype=torch.float16, variant="fp16")
pipe.to("cuda") pipe.to("cuda")
# pipe.vae = AutoencoderTiny.from_pretrained('madebyollin/taesdxl', torch_device='cuda', torch_dtype=torch.float16) pipe.vae = AutoencoderTiny.from_pretrained('madebyollin/taesdxl', torch_device='cuda', torch_dtype=torch.float16)
# pipe.vae = pipe.vae.cuda() pipe.vae = pipe.vae.cuda()
dh = DiffusersHolder(pipe) dh = DiffusersHolder(pipe)
# %% Next let's set up all parameters # %% Next let's set up all parameters
depth_strength = 0.5 # Specifies how deep (in terms of diffusion iterations the first branching happens)
t_compute_max_allowed = 5 # Determines the quality of the transition in terms of compute time you grant it
num_inference_steps = 4
size_output = (512, 512) size_output = (512, 512)
prompt1 = "underwater landscape, fish, und the sea, incredible detail, high resolution" prompt1 = "underwater landscape, fish, und the sea, incredible detail, high resolution"
prompt2 = "rendering of an alien planet, strange plants, strange creatures, surreal" prompt2 = "rendering of an alien planet, strange plants, strange creatures, surreal"
negative_prompt = "blurry, ugly, pale" # Optional negative_prompt = "blurry, ugly, pale" # Optional
fp_movie = 'movie_example1.mp4'
duration_transition = 12 # In seconds duration_transition = 12 # In seconds
# Spawn latent blending # Spawn latent blending
@ -865,31 +877,24 @@ if __name__ == "__main__":
lb.set_prompt2(prompt2) lb.set_prompt2(prompt2)
lb.set_dimensions(size_output) lb.set_dimensions(size_output)
lb.set_negative_prompt(negative_prompt) lb.set_negative_prompt(negative_prompt)
lb.set_guidance_scale(0)
lb.branch1_crossfeed_power = 0.0
lb.branch1_crossfeed_range = 0.6
lb.branch1_crossfeed_decay = 0.99
lb.parental_crossfeed_power = 1.0
lb.parental_crossfeed_power_decay = 1.0
lb.parental_crossfeed_range = 1.0
# Run latent blending # Run latent blending
lb.run_transition( lb.run_transition(fixed_seeds=[420, 421])
depth_strength=depth_strength,
num_inference_steps=num_inference_steps,
t_compute_max_allowed=t_compute_max_allowed)
# Save movie # Save movie
fp_movie = f'test.mp4'
lb.write_movie_transition(fp_movie, duration_transition) lb.write_movie_transition(fp_movie, duration_transition)
#%% #%%
""" """
checkout sizes
checkout good tree for num inference steps checkout good tree for num inference steps
checkout that good nmb inference step given checkout that good nmb inference step given
timing1: dt_per_diff rename and fix (first time run is super slow)
timing2: measure time for decoding
""" """