<h1 id="important-please-read-before-reporting-character-lora-issues">IMPORTANT — PLEASE READ BEFORE REPORTING CHARACTER LoRA ISSUES</h1>The Krea2T Character LoRA — Image Only loader does NOT behave like a normal LoRA loader, and it is not intended to.Comparing both loaders at <code>1.00</code> strength and assuming they should produce identical character likeness is an invalid comparison.This node intentionally applies the character LoRA only to the image-token rows inside Krea2’s shared DiT blocks. It does not directly apply the character LoRA to the text-token rows.Text and image still interact normally through Krea2’s native attention; the character LoRA itself is simply prevented from directly modifying the text side.That behavior is intentional.Strength Is Not 1:1 With a Normal LoRA LoaderBecause the character LoRA is being applied through a more restricted mechanism, <code>1.00</code> on this loader is not equivalent to <code>1.00</code> on a conventional LoRA loader.If your character likeness becomes weaker, increase the character LoRA strength above <code>1.00</code> until the desired likeness is recovered.The node allows strengths above <code>1.00</code> for exactly this reason.Why the Loader Works This WayThe refusal-reduction LoRA and the character LoRA have different jobs.The refusal-reduction LoRA is intended to alter the model behavior responsible for refusal.The character LoRA should provide the character appearance without unnecessarily becoming part of the prompt-interpretation path that the refusal-reduction LoRA is working against.The entire point of this loader is therefore to apply the character LoRA as image influence rather than as a whole-entity influence across the full shared path.Please Keep This in Mind Before Reporting an Issue- Do not assume equal strength values between this loader and a normal LoRA loader represent equal influence- Weaker likeness at <code>1.00</code> does not mean the loader is broken. Increase the character strength.- This loader intentionally prevents direct character-LoRA contribution to text-token rows.- This does not mean the character LoRA is “not working.”- A difference in likeness does not prove that the likeness is located in the text encoder.- If you specifically want the character LoRA applied conventionally across the normal model path, **use your normal LoRA loader instead.Intended SeparationThe design is simple:Refusal-reduction LoRA → deals with refusal behaviorCharacter LoRA → supplies the visual character informationThe goal is to keep the character LoRA from unnecessarily interfering with the interpretation path that the refusal-reduction LoRA is designed to modify.Please understand what the node is designed to do before concluding that the behavior is a bug.Now supports LoKr as well.<hr /><span style="background-color:rgb(26, 27, 30);color:rgb(64, 192, 87);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">sample workflow (Updated with new Loader) <span style="background-color:rgb(26, 27, 30);color:rgb(193, 194, 197);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">: <a target="_blank" rel="ugc" href="https://github.com/capitan01R/ComfyUI-Krea2T-Enhancer/blob/main/workflows/character_LoRA_wf_updated.json"><span style="background-color:rgb(26, 27, 30);color:rgb(76, 110, 245);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">https://github.com/capitan01R/ComfyUI-Krea2T-Enhancer/blob/main/workflows/character_LoRA_wf_updated.json</a><h1 id="krea2-refusal-reduction-v2-and-v2-full-rank">Krea2 Refusal-Reduction — v2 and v2 Full Rank</h1>Okay, I am adding two versions this time: v2 and v2 Full Rank.The goal is still the same: reduce refusal behavior while preserving as much of the model's existing visual knowledge and prompt adherence as possible. Getting a concept to appear initially is only part of it. I also want it to hold together through the denoise without drifting into something else or ending up with those weird textures.There is also an important change to the setup if you use character LoRAs. Please read that part before loading everything together, because conflicts between them have been a recurring problem.<hr />The Two Versionsv2File: <code>refusal_reduction_v2.safetensors</code>This is the smaller rank-64 version, approximately 30 MB. It includes updates to the four TextFusion blocks, the tap projector, and the two external text MLP layers that prepare the conditioning for the image transformer.v2 Full RankFile: <code>refusal_reduction_v2_full_rank.safetensors</code>This is the larger full-rank export, approximately 2.6 GB. It covers the same main conditioning components and also includes selected normalization adjustments.The matrix updates are stored at their full available rank in FP32, which accounts for the much larger file. “Full Rank” describes how those updates are represented; the changes are still confined to the text-conditioning components covered by this release.Load one version only. Start at strength <code>1.00</code>.<hr />What These Versions ChangeKrea2 takes several representations of your prompt from its text encoder and processes them before they reach the image transformer. These releases modify that processing, with the aim of making the model less likely to suppress the requested concept.Both versions include:• The two TextFusion layerwise blocks.• The two TextFusion refiner blocks.• The projector that combines the twelve text-encoder representations.• The two external text MLP layers that feed the resulting conditioning into the image transformer.The earlier description saying the projector and external text MLPs were excluded does not apply to these two files.Neither file contains updates to the image-transformer blocks. They change the conditioning those blocks receive, which can still change the resulting image substantially.The purpose remains refusal reduction. I am not aiming to impose a character, style, or preferred aesthetic with these releases.<hr />Using a Character LoRA? This Loader Is RequiredIf you combine either version with a character LoRA, use <a target="_blank" rel="ugc" href="https://github.com/capitan01R/ComfyUI-Krea2T-Enhancer/tree/main#character-lora--image-only">Krea2T Character LoRA — Image Only</a> to load the character LoRA.One thing I have repeatedly run into is a character LoRA pulling an otherwise working setup away from the requested concept. You get the likeness, but other parts of the prompt start behaving differently or become less reliable.Simply excluding weights from the dedicated text-processing layers does not fully separate their influence. Krea2 also has shared layers that process both the prompt and the image. A character LoRA applied normally to those layers affects both.That is why I added this loader. It filters the supported character weights and applies their direct updates only to image tokens. The refusal-reduction LoRA stays active through its own loader.Text and image still interact through the model's normal attention, so some interaction remains. The loader targets that direct overlap rather than treating every conflict as a reason to keep increasing LoRA strength.Connect It Like ThisKrea2 model ↓ Regular LoRA loader — refusal_reduction_v2 OR refusal_reduction_v2_full_rank ↓ Krea2T Character LoRA — Image Only — your character LoRA ↓ Sampler• Load the refusal-reduction LoRA through your regular model LoRA loader.• Load the character LoRA through the Image Only loader. Replace its previous loader; do not load the same character twice.• Do not put the refusal-reduction LoRA into the Image Only loader. Its filtering excludes the conditioning weights these releases use.• If you are not using a character LoRA, connect the refusal-reduction loader directly to the rest of your workflow.<hr />Still a Work in ProgressI am continuing to check how consistently this holds up across prompts, seeds, and different LoRA combinations. The objective is useful refusal reduction with the concept intact through generation. I am not calling every changed image a success.<a target="_blank" rel="ugc" href="https://buymeacoffee.com/capitan01r">Buy me a coffee</a> :)<hr />Responsibility and Output DisclaimerThis LoRA modifies the model's text-conditioning behavior but does not determine, authorize, endorse, supervise, or control the images produced by individual users, prompts, workflows, checkpoints, samplers, or third-party software.Generated outputs remain dependent on the base model, user-provided conditioning, inference configuration, and the surrounding generation pipeline. The author assumes no responsibility or liability for what users choose to generate, how generated material is used, or whether any output complies with applicable laws, platform rules, licensing terms, or third-party rights.Users are solely responsible for their prompts, generated outputs, distribution decisions, and use of this LoRA.

Krea2 TextFusion Refusal-Reduction LoRA (Updated)

refusal_reduction_v2_full_rank.safetensors

Krea2 TextFusion Refusal-Reduction LoRA

v2.0 Full Rank

v2.0

v1.0

A style LoRA for Krea 2 that brings the look of modern, high-budget anime and key visuals: clean, consistent lineart, crisp cel shading with hard-edged shadows, vibrant colours and detailed eyes. It was trained on a large dataset of anime key visuals and high resolution digital anime illustrations.<ul><li>no AI images in the dataset</li><li>no triggerword</li><li>Detailed natural-language captions in English, no tags.</li></ul>In my first testing the LoRA made also non Anime checkpoints output anime style without prompting for it but it seems this is rather rare so, you should still prompt the model for anime style and the LoRA will then steer that to the modern key visual style.<h3 id="v2"> V2</h3>While i liked the first version i went again through my dataset and decided that the style of some images is too different from the general style of the dataset, so i removed almost all digital anime wallpaper illustrations and some landscape scenery illustrations and added more key visuals and some new anime "no human sceneries", so the dataset has a more consistent and uniform style. If the resulting LoRA is better is probably a matter of taste, but at least i can say the dataset is better. V2 gives lighter and cleaner skin which i like.My showcase images contain my ComfyUI workflow (if you want to get the exact style i showed off in the showcase, there you have the exact settings i used).I trained 3000 steps but went with the 2200, if you think the LoRA is too weak i can upload a stronger version.<h3 id="v3">V3</h3>The goal was to get the intense eyes and vibrant colours from V1 while keeping the lighter and cleaner skin from V2. I also replaced the no human sceneries from my dataset with new ones, i now used Screenshots from high budget anime for no human sceneries (no screenshots for characters, there only key visuals and offical arts were used), since i did not like the sceneries i had, i just used them because it was the best i found. I also enhanced the captions. I also added more images where the characters smile with showing their teeth. I would say the resulting LoRA reached the goal to get more intense eyes and also the prompt adherence is better.

Krea 2 Anime Style

Krea2_TextFusion_Refusal_Reduction.safetensors

refusal_reduction_v2.safetensors

Krea2_Anime_Style_V2.safetensors

v3.0

<h1 id="krea2-textfusion-refusal-reduction-by-capitan01r">Krea2 TextFusion Refusal-Reduction [by Capitan01R]</h1>No idea what happened with Capitan01RI will happily take the model down if <a target="_blank" rel="ugc" href="https://civarchive.com/user/Capitan01R">Capitan01R</a> decides to come back. Until then yellow buzz generated with this gets tipped to him. If he deletes his account I'll throw the buzz to some random dude.But this is a great LoRA and it should be available to the community.Update: Capitan01R politely declined the yellow buzz, so it will be random users.<hr /># Krea2 TextFusion Refusal-Reduction LoRAThis is a dedicated rank-64 LoRA trained with one narrow objective: reduce Krea2’s learned refusal and restriction behavior while preserving as much of the base model’s existing visual knowledge and prompt behavior as possible.This is not a concept, character, style, anatomy, or aesthetic LoRA. It was not trained to add new visual concepts, reproduce a training-image set, or impose a preferred look. Its training objective was exclusively to push the model away from refusal behavior so that concepts already represented within the base model are less likely to be suppressed during text conditioning.## What It ChangesKrea2 receives multiple hidden-state taps from its Qwen-VL text encoder and processes them through TextFusion before sending the resulting conditioning into the image transformer. This LoRA applies learned low-rank residuals only to the attention and internal MLP projections within:- txtfusion.layerwise_blocks.0- txtfusion.layerwise_blocks.1- txtfusion.refiner_blocks.0- txtfusion.refiner_blocks.1The release version deliberately contains no adapter for:- The TextFusion 1 × 12 tap projector- The external/general txtmlp- The image-transformer blocks or any other image-generation layers## How It WorksThe base checkpoint is never overwritten. For every targeted linear layer, LoRA adds a learned rank-64 residual to the original weight during inference:W_effective = W_base + strength × ΔWThis changes how existing Qwen-VL text features are routed, gated, transported, and refined through TextFusion before they reach Krea2’s untouched image transformer. In other words, the LoRA is intended to improve access to visual knowledge already present in the base model, not inject a new concept or replace the model’s learned visual representations.This release targets the specific TextFusion route isolated through layer-by-layer ablation instead of broadly amplifying activations or directly altering the projector.## UsageUse strength of "1.00"Responsibility and Output DisclaimerThis LoRA modifies the model’s text-conditioning behavior but does not determine, authorize, endorse, supervise, or control the images produced by individual users, prompts, workflows, checkpoints, samplers, or third-party software.Generated outputs remain dependent on the base model, user-provided conditioning, inference configuration, and the surrounding generation pipeline. The author assumes no responsibility or liability for what users choose to generate, how generated material is used, or whether any output complies with applicable laws, platform rules, licensing terms, or third-party rights.Users are solely responsible for their prompts, generated outputs, distribution decisions, and use of this LoRA.

Capitan01R's Krea2 TextFusion Refusal-Reduction

Capitan01R_Krea2_TextFusion_Refusal_Reduction_v2_FullRank.safetensors

v2_Full_Rank

Files with SHA256: 999c6bc160d6c323309395a6abea416883074b0cf429511ee33cba28ce9a8d89