jinx/ComfyUI
1
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2025-08-02 23:14:49 +08:00
Code Issues Activity

Implement support for taef1 latent previews (#4409)

Browse Source 
* add taef1 handling to several places

* remove guess_latent_channels and add latent_channels info directly to flux model

* remove TODO

* fix numbers
This commit is contained in:
Matthew Turnshek
2024-08-16 12:53:13 -04:00
committed by GitHub
parent 05a9f3faa1
commit 1770fc77ed
2 changed files with 14 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
comfy/latent_formats.py
View File

@@ -141,6 +141,7 @@ class StableAudio1(LatentFormat):
latent_channels = 64
class Flux(SD3):
latent_channels = 16
def __init__(self):
self.scale_factor = 0.3611
self.shift_factor = 0.1159
@@ -162,6 +163,7 @@ class Flux(SD3):
[-0.0005, -0.0530, -0.0020],
[-0.1273, -0.0932, -0.0680]
]
self.taesd_decoder_name = "taef1_decoder"
def process_in(self, latent):
return (latent - self.shift_factor) * self.scale_factor