A simple workflow for Klein family models in the first row was developed for 9b and 4b distilled versions, but it should work for base models with minimal changes (such as replacing ConditioningZeroOut with negative conditioning).
For legacy mode, both ComfyUI preview previews and the mute/bypass toggle within subgraphs are broken again. I recommend launching ComfyUI using either
--front-end-version Comfy-Org/[email protected] or
--front-end-version Comfy-Org/[email protected]The workflow is for legacy mode. I am not planning to make it useful or even test for Nodes 2.0 mode.
🚩vX Latent This is an experimental versions of the workflow. It’s not intended for general use (though it might still work as a regular one🟢). The main goal here is to slow down degradation when editing images individually by one-by-one editioning. Typically, using VAE decode artifacts appears after 3–5 edits, which can be addressed by simply removing the latent encoding/decoding steps via two nodes: Latent Loader and Latent Save. This approach fully bypasses VAE entirely for preview purposes—only to use it during editing previews.
I implemented a chain of K-samplers and used a specific schema; while there are no visible artifacts, over time the colors fade slightly, but this is still better than the standard Edit mode. If you try it out, please provide feedback about this workflow (whether positive or negative). I recognize that other methods exist for restoring images, but they’re much more complicated—so I’m exploring this approach first.
At least for now, changing the size in Latent Edit mode isn’t available. It’s possible, but it comes with many limitations; I didn’t find the way I like. So you’ll have the same size as what was saved (.latent file).
Description
v5.6:
Added the SeedVR2 upscaler.
Added a comparer preview for SeedVR2.
Changed the README note.
Changed the links note.
Made some visual changes.
v5.7:
HotFix. Removed PlaySoundKJ node. It's just stopd working with no reason.
FAQ
Comments (4)
Thank you for the very nice workflow. There's a problem though. For some strange reason I cannot add a LoRA via LoRA loader. I see loRAs that existed before I added this workflow and whatever I do I cannot add new one, no .json file is created and when I add .json file myself I still cannot see it in a loader menu. From the other hand, I cannot remove a LoRA, I still see it in the menu after removing a LoRA file and a .json. Am I doing something wrong? Maybe there's some reindex procedure? Could you please clarify, Thanks.
@sergeyouknw I'm not sure I understand you correctly. It seems you're having trouble adding LORA to the LORA Manager, right? Is this behavior limited to just the LORA Loader in a new Canvas, or does it happen specifically in my workflow?
@DemonAlone , yes, it seems it happens only in your workflow. I currently have only two active workflows in my ComfyUI setup, one for LTX video and the one of yours. LTX one has its own LoRA manager where I can directly add LoRAs (pressing a button adds new line where I can directly browse and select a LoRA file), and in your workflow i start entering "lora" in the field and the manager should show the list of all LoRA files available in the dedicated standard folder and I always see only the 5 LoRAs added previously. And whatever I do, I cannot change that list (not remove, nor add).
@sergeyouknw , I am afraid I can’t do anything with it. I didn’t have any “influence” on other nodes; at least that wasn’t part of my plan. All I can offer is to try the oldest version of that workflow or any other workflows using lora manager (z-image, qwen, wan2.2, krea, sdxl-like), and check if the behavior persists. If so, that’s bad news. Another thing you could try is deleting group by group to see which node might be causing the corruption. If only lora manager remains in the workflow with no change, then there may be a problem elsewhere. But if I do it and identify which node caused the issue (if that’s the reason), please let me know – perhaps that will clarify what went wrong.