Integrate tiled upscale into Continuum second pass
This commit is contained in:
@@ -17,6 +17,16 @@ context-to-conditioning adaptation as **H3 Continuum Conditioning Bridge V3.5**.
|
||||
The separate `conditioning` input is optional for advanced legacy graphs; it is
|
||||
not needed in the normal Continuum wiring.
|
||||
|
||||
The recommended production wiring is now the **H3 Continuum Second Pass V3.5**
|
||||
node with `tiled_upscale = true`. That node receives the main sampler's video
|
||||
and audio latent lists plus its `refine_context` and `assembly_plan`, rebuilds a
|
||||
complete AV latent for each physical group, and runs the native tiled upscale
|
||||
one group at a time. It returns the upscaled video latent list while preserving
|
||||
the first-pass audio latent list. Connect the latent-upscale and spatial-split
|
||||
parameter nodes to its optional tiled inputs; its sampler, scheduler, steps,
|
||||
denoise, CFG and noise seed are independent from the ordinary Second Pass
|
||||
settings.
|
||||
|
||||
The node is an adapter around ComfyUI's native MiniMax H3 tiled-upscale
|
||||
implementation; it does not include model weights or vendor the upstream
|
||||
implementation. Install this repository as a normal custom-node folder and
|
||||
|
||||
Reference in New Issue
Block a user