Add FLUX.2-klein image edit workflows
This commit is contained in:
@@ -12,3 +12,12 @@ Recommended subcategories to watch for:
|
|||||||
- LoRA showcase packs that include a practical workflow
|
- LoRA showcase packs that include a practical workflow
|
||||||
|
|
||||||
Store each workflow in its own subfolder.
|
Store each workflow in its own subfolder.
|
||||||
|
|
||||||
|
## Current note
|
||||||
|
|
||||||
|
The strongest trustworthy starting point right now is the official Comfy workflow template set from the `docs.comfy.org` `FLUX.2-klein` guide, especially the image-edit templates.
|
||||||
|
|
||||||
|
For outpainting specifically:
|
||||||
|
|
||||||
|
- start with `image-edit-4b-distilled/` if you want the lightest and fastest first try
|
||||||
|
- move to `image-edit-9b-base/` or `image-edit-9b-distilled/` if quality is strong enough to justify the heavier setup
|
||||||
|
|||||||
@@ -0,0 +1,46 @@
|
|||||||
|
# FLUX.2-klein Image Edit 4B Base
|
||||||
|
|
||||||
|
## Summary
|
||||||
|
|
||||||
|
Official `FLUX.2-klein` `4B base` image-edit workflow from the Comfy workflow templates. Good for general editing tasks when you want the more flexible base model rather than the faster distilled path.
|
||||||
|
|
||||||
|
## Model Family
|
||||||
|
|
||||||
|
- `flux2klein`
|
||||||
|
|
||||||
|
## Status
|
||||||
|
|
||||||
|
- `imported-only`
|
||||||
|
|
||||||
|
## Source
|
||||||
|
|
||||||
|
- Official guide: <https://docs.comfy.org/tutorials/flux/flux-2-klein>
|
||||||
|
- Workflow file: <https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/image_flux2_klein_image_edit_4b_base.json>
|
||||||
|
- Original publisher: `Comfy-Org`
|
||||||
|
- Date imported: `2026-08-08`
|
||||||
|
|
||||||
|
## Developer notes
|
||||||
|
|
||||||
|
- This is one of the official workflow templates linked directly from the Comfy docs.
|
||||||
|
- The guide positions `FLUX.2-klein` as a unified text-and-edit model family for interactive editing, object changes, multi-reference work, and iterative edits.
|
||||||
|
- The workflow includes a documentation note listing the expected model filenames and storage locations.
|
||||||
|
|
||||||
|
## Our notes
|
||||||
|
|
||||||
|
- Best use: trustworthy baseline for `FLUX.2-klein` editing without betting on a random community graph.
|
||||||
|
- This is probably the safer "quality over speed" 4B edit starting point.
|
||||||
|
- It is not labeled outpainting explicitly, but it is the same family of edit workflow you would use as an outpainting base after expanding the canvas / source image setup in Comfy.
|
||||||
|
|
||||||
|
## Required custom nodes
|
||||||
|
|
||||||
|
- none beyond current official Comfy support for the `FLUX.2-klein` template nodes
|
||||||
|
|
||||||
|
## Required model files
|
||||||
|
|
||||||
|
- `models/diffusion_models/flux-2-klein-base-4b-fp8.safetensors`
|
||||||
|
- `models/text_encoders/qwen_3_4b.safetensors`
|
||||||
|
- `models/vae/full_encoder_small_decoder.safetensors`
|
||||||
|
|
||||||
|
## Notes
|
||||||
|
|
||||||
|
- If you mainly care about speed for outpainting experiments, use the `4B distilled` entry first instead.
|
||||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,48 @@
|
|||||||
|
# FLUX.2-klein Image Edit 4B Distilled
|
||||||
|
|
||||||
|
## Summary
|
||||||
|
|
||||||
|
Official `FLUX.2-klein` `4B distilled` image-edit workflow from Comfy's template set. This is the best outpainting starter in the repo right now for `FLUX.2-klein` because it is official, current, and light enough to iterate quickly.
|
||||||
|
|
||||||
|
## Model Family
|
||||||
|
|
||||||
|
- `flux2klein`
|
||||||
|
|
||||||
|
## Status
|
||||||
|
|
||||||
|
- `imported-only`
|
||||||
|
|
||||||
|
## Source
|
||||||
|
|
||||||
|
- Official guide: <https://docs.comfy.org/tutorials/flux/flux-2-klein>
|
||||||
|
- Workflow file: <https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/image_flux2_klein_image_edit_4b_distilled.json>
|
||||||
|
- Original publisher: `Comfy-Org`
|
||||||
|
- Date imported: `2026-08-08`
|
||||||
|
|
||||||
|
## Developer notes
|
||||||
|
|
||||||
|
- This is the official fast distilled 4B image-edit template referenced by the Comfy docs.
|
||||||
|
- The docs describe `FLUX.2-klein` as supporting style transforms, semantic edits, object replacement/removal, multi-reference composition, and iterative edits.
|
||||||
|
- The `4B distilled` path is presented as the speed-first option.
|
||||||
|
|
||||||
|
## Our notes
|
||||||
|
|
||||||
|
- This is the one I would try first for outpainting.
|
||||||
|
- Reason: it is official, lower-risk, and should be much easier to iterate on than a heavier 9B setup.
|
||||||
|
- Even though the workflow is labeled image edit rather than outpainting, the same template is the most credible base for outpainting in `FLUX.2-klein` right now.
|
||||||
|
- If this works but feels too weak or too compressed, step up to `9B distilled` or `9B base`.
|
||||||
|
|
||||||
|
## Required custom nodes
|
||||||
|
|
||||||
|
- none beyond current official Comfy support for the `FLUX.2-klein` template nodes
|
||||||
|
|
||||||
|
## Required model files
|
||||||
|
|
||||||
|
- `models/diffusion_models/flux-2-klein-4b-fp8.safetensors`
|
||||||
|
- `models/text_encoders/qwen_3_4b.safetensors`
|
||||||
|
- `models/vae/full_encoder_small_decoder.safetensors`
|
||||||
|
|
||||||
|
## Notes
|
||||||
|
|
||||||
|
- Best current `FLUX.2-klein` outpainting starter in this repo.
|
||||||
|
- When you test it, we may want to update the model paths or filenames inside the JSON if your local Comfy setup uses different names.
|
||||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,45 @@
|
|||||||
|
# FLUX.2-klein Image Edit 9B Base
|
||||||
|
|
||||||
|
## Summary
|
||||||
|
|
||||||
|
Official `FLUX.2-klein` `9B base` image-edit workflow from Comfy's template set. Heavier than the 4B options, but likely the better choice if you want maximum flexibility from the official `klein` edit stack.
|
||||||
|
|
||||||
|
## Model Family
|
||||||
|
|
||||||
|
- `flux2klein`
|
||||||
|
|
||||||
|
## Status
|
||||||
|
|
||||||
|
- `imported-only`
|
||||||
|
|
||||||
|
## Source
|
||||||
|
|
||||||
|
- Official guide: <https://docs.comfy.org/tutorials/flux/flux-2-klein>
|
||||||
|
- Workflow file: <https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/image_flux2_klein_image_edit_9b_base.json>
|
||||||
|
- Original publisher: `Comfy-Org`
|
||||||
|
- Date imported: `2026-08-08`
|
||||||
|
|
||||||
|
## Developer notes
|
||||||
|
|
||||||
|
- This is the official `9B base` edit template from Comfy's published workflow list.
|
||||||
|
- The docs point users to the BFL repo for acceptance and download of the 9B diffusion weights.
|
||||||
|
|
||||||
|
## Our notes
|
||||||
|
|
||||||
|
- Use this when the `4B` variants are good enough to prove the workflow direction, but you want a stronger ceiling.
|
||||||
|
- More of a second-step workflow than a first try, especially if your box is tight on VRAM.
|
||||||
|
- Also a plausible outpainting path once the lighter template is validated.
|
||||||
|
|
||||||
|
## Required custom nodes
|
||||||
|
|
||||||
|
- none beyond current official Comfy support for the `FLUX.2-klein` template nodes
|
||||||
|
|
||||||
|
## Required model files
|
||||||
|
|
||||||
|
- `models/diffusion_models/flux-2-klein-base-9b-fp8.safetensors`
|
||||||
|
- `models/text_encoders/qwen_3_8b_fp8mixed.safetensors`
|
||||||
|
- `models/vae/flux2-vae.safetensors`
|
||||||
|
|
||||||
|
## Notes
|
||||||
|
|
||||||
|
- Better treated as a quality-upgrade path, not the first one to debug.
|
||||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,45 @@
|
|||||||
|
# FLUX.2-klein Image Edit 9B Distilled
|
||||||
|
|
||||||
|
## Summary
|
||||||
|
|
||||||
|
Official `FLUX.2-klein` `9B distilled` image-edit workflow from Comfy's template set. Faster than `9B base` while still targeting the larger model family.
|
||||||
|
|
||||||
|
## Model Family
|
||||||
|
|
||||||
|
- `flux2klein`
|
||||||
|
|
||||||
|
## Status
|
||||||
|
|
||||||
|
- `imported-only`
|
||||||
|
|
||||||
|
## Source
|
||||||
|
|
||||||
|
- Official guide: <https://docs.comfy.org/tutorials/flux/flux-2-klein>
|
||||||
|
- Workflow file: <https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/image_flux2_klein_image_edit_9b_distilled.json>
|
||||||
|
- Original publisher: `Comfy-Org`
|
||||||
|
- Date imported: `2026-08-08`
|
||||||
|
|
||||||
|
## Developer notes
|
||||||
|
|
||||||
|
- This is the official fast `9B distilled` image-edit template linked from the Comfy docs page.
|
||||||
|
- The docs position the distilled variants as the speed-first route inside the family.
|
||||||
|
|
||||||
|
## Our notes
|
||||||
|
|
||||||
|
- This is probably the best "bigger model, still practical" upgrade path after testing `4B distilled`.
|
||||||
|
- Good candidate if you like the edit behavior of the smaller workflow but want a bit more headroom without immediately jumping to the full base path.
|
||||||
|
- Also suitable as an outpainting follow-up if the 4B result is too soft.
|
||||||
|
|
||||||
|
## Required custom nodes
|
||||||
|
|
||||||
|
- none beyond current official Comfy support for the `FLUX.2-klein` template nodes
|
||||||
|
|
||||||
|
## Required model files
|
||||||
|
|
||||||
|
- `models/diffusion_models/flux-2-klein-9b-fp8.safetensors`
|
||||||
|
- `models/text_encoders/qwen_3_8b_fp8mixed.safetensors`
|
||||||
|
- `models/vae/flux2-vae.safetensors`
|
||||||
|
|
||||||
|
## Notes
|
||||||
|
|
||||||
|
- Best upgrade path after the `4B distilled` template if you want to stay in the same workflow shape.
|
||||||
File diff suppressed because it is too large
Load Diff
Reference in New Issue
Block a user