nodes/Kosinkadink-ComfyUI-AnimateDiff-Evolved
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

NaiveReuse is now functional for Uniform Context Options and relative fuse_method

Browse Source
This commit is contained in:
Jedrzej Kosinski
2024-07-29 01:35:12 -05:00
parent d1fc329c61
commit 1c68c56809
1 changed files with 19 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
animatediff/sampling.py
+19 -12
View File
@@ -820,7 +820,12 @@ def sliding_calc_conds_batch(model, conds, x_in: Tensor, timestep, model_options
# prepare final conds, out_counts, and biases
conds_final = [torch.zeros_like(x_in) for _ in conds]
counts_final = [torch.zeros((x_in.shape[0], 1, 1, 1), device=x_in.device) for _ in conds]
if ADGS.params.context_options.fuse_method == ContextFuseMethod.RELATIVE:
# counts_final not used for RELATIVE fuse_method
counts_final = [torch.ones((x_in.shape[0], 1, 1, 1), device=x_in.device) for _ in conds]
else:
# default counts_final initialization
counts_final = [torch.zeros((x_in.shape[0], 1, 1, 1), device=x_in.device) for _ in conds]
biases_final = [([0.0] * x_in.shape[0]) for _ in conds]
CONTEXTREF_CONTROL_LIST_ALL = "contextref_control_list_all"
@@ -910,22 +915,24 @@ def sliding_calc_conds_batch(model, conds, x_in: Tensor, timestep, model_options
model_options["transformer_options"][CONTEXTREF_CLEAN_FUNC]()
context_ref_injector.restore()
# handle NaiveReuse
if cached_naive_conds is not None:
start_idx = cached_naive_ctx_idxs[0]
for z in range(0, ADGS.params.full_length, len(cached_naive_ctx_idxs)):
for i in range(len(cached_naive_conds)):
# get the 'true' idxs of this window
new_ctx_idxs = [(zz+start_idx) % ADGS.params.full_length for zz in list(range(z, z+len(cached_naive_ctx_idxs))) if zz < ADGS.params.full_length]
# make sure when getting cached_naive idxs, they are adjusted for actual length leftover length
adjusted_naive_ctx_idxs = cached_naive_ctx_idxs[:len(new_ctx_idxs)]
weighted_mean = ADGS.params.context_options.extras.naive_reuse.get_effective_weighted_mean(x_in, new_ctx_idxs)
conds_final[i][new_ctx_idxs] = (weighted_mean * (cached_naive_conds[i][adjusted_naive_ctx_idxs]*counts_final[i][new_ctx_idxs])) + ((1.-weighted_mean) * conds_final[i][new_ctx_idxs])
del cached_naive_conds
if ADGS.params.context_options.fuse_method == ContextFuseMethod.RELATIVE:
# already normalized, so return as is
del counts_final
return conds_final
else:
if cached_naive_conds is not None:
#start_idx = cached_naive_ctx_idxs[-1] + 1
start_idx = cached_naive_ctx_idxs[0]
for z in range(start_idx, ADGS.params.full_length, len(cached_naive_ctx_idxs)):
for i in range(len(cached_naive_conds)):
new_ctx_idxs = [zz for zz in list(range(z, z+len(cached_naive_ctx_idxs))) if zz < ADGS.params.full_length]
# make sure when getting cached_naive idxs, they are adjusted for actual length leftover length
adjusted_cnaive_ctx_idxs = cached_naive_ctx_idxs[:len(new_ctx_idxs)]
weighted_mean = ADGS.params.context_options.extras.naive_reuse.get_effective_weighted_mean(x_in, new_ctx_idxs)
conds_final[i][new_ctx_idxs] = (weighted_mean * (cached_naive_conds[i][adjusted_cnaive_ctx_idxs]*counts_final[i][new_ctx_idxs])) + ((1.-weighted_mean) * conds_final[i][new_ctx_idxs])
del cached_naive_conds
# normalize conds via division by context usage counts
for i in range(len(conds_final)):
conds_final[i] /= counts_final[i]