Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-19 09:14:35 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix(ltx): handle PIL list output from refine path in interpolation wiring

Browse Source 
LTX refine pipe uses output_type='pil' so result.frames[0] returns a list, not a 5-D tensor. Convert via images_to_tensor before the helper sees it, mirroring what save_video already does for the same input.
This commit is contained in:
CalamitousFelicitousness
2026-04-27 01:13:47 +01:00
parent 7da3a15305
commit b48bf5236b
1 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/ltx/ltx_process.py
+6 -2
View File
@@ -563,10 +563,14 @@ def run_ltx(task_id,
except Exception:
aac_sample_rate = 24000
if mp4_interpolate > 0:
if mp4_interpolate > 0 and pixels is not None:
p.video_interpolate = mp4_interpolate
from modules.processing_video import apply_video_interpolation
# pixels is 5-D (N,C,T,H,W); RIFE needs 4-D (T,C,H,W) in [0,1]
# refine path returns PIL list (output_type='pil'); decode path returns 5-D tensor
if isinstance(pixels, list) and len(pixels) > 0 and isinstance(pixels[0], Image.Image):
from modules.video_models.video_save import images_to_tensor
pixels = images_to_tensor(pixels)
# pixels is 5-D (N,C,T,H,W) in [-1,1]; RIFE needs 4-D (T,C,H,W) in [0,1]
x = pixels.squeeze(0).permute(1, 0, 2, 3)
x = (x.clamp(-1., 1.) + 1.0) * 0.5
x = apply_video_interpolation(p, x, count=mp4_interpolate)