LTXV Chunk FeedForward : Question ?

[monnky]

Hi @kijai

will this node do the upscaler sampling in chunks ? i mean in batches like 76 images 126 images (25fps) ?

[kijai]

No, it just reduces the model's peak VRAM use by chunking the function that uses most VRAM. So just saves you some VRAM.

[monnky]

@kijai thanks for the info, is there any way, we can do the spatial upscaler sampling in batches, like meta batch in video combine node ??

[kijai]

@kijai thanks for the info, is there any way, we can do the spatial upscaler sampling in batches, like meta batch in video combine node ??

Check the LTXNodes v2v detailer workflow, it uses their looping node to do something like that.

[monnky]

@kijai thanks for all the help.

[monnky]

@kijai its okay, will try to find any other solution for batching.