<h3 id="if-you-enjoy-my-work-and-would-like-to-support-me-consider-buying-me-a-coffee.">❤️ If you enjoy my work and would like to support me, consider buying me a <a target="_blank" rel="ugc" href="https://ko-fi.com/latentheart#linkModal">☕ Coffee</a>.</h3> This release provides a ComfyUI-compatible safetensors package of <a target="_blank" rel="ugc" href="https://huggingface.co/huihui-ai/Huihui-Qwen3-VL-4B-Instruct-abliterated">Huihui-Qwen3-VL-4B-Instruct-abliterated</a>, converted into the format expected by the Krea 2 text encoder loader.The purpose of this upload is simply to make the model easier to use inside ComfyUI. The original model, the abliteration work, and the conversion tools were all created by other developers. I only followed their work to produce a ready-to-use ComfyUI package.<h2 id="what-is-this">What is this?</h2>This is a vision-language version of Qwen3-VL 4B that has been abliterated and converted into a single ComfyUI-compatible safetensors file.It can be used as a Krea 2 text encoder inside ComfyUI while preserving the model's vision capabilities, making it suitable for workflows that use image-aware prompt enhancement.A quick note about abliterated text encodersOne point that often causes confusion (especially for people who are new to ComfyUI workflows) is the role of the text encoder.If you're using Qwen3-VL only as a text encoder (for example, through Krea 2's CLIPLoader), the model is simply converting your prompt into embeddings. It is processing an input, not generating a text response, so it won't explicitly block you or spit out a refusal message.However, safety tuning still affects the math. When a standard model processes an "unsafe" prompt, its internal embeddings often collapse into a generic, flattened state rather than retaining the rich details of your description. This can subtly degrade your image generation. An abliterated (or "heretic") model removes this refusal behavior, ensuring your unrestricted prompts remain highly detailed when passed to the diffusion model.This release exists because it provides a ComfyUI-compatible, vision-capable Qwen3-VL text encoder in a convenient safetensors format. The fact that the underlying model is abliterated is a bonus for unrestricted image generation for NSFW content, but is absolutely essential for workflows that also make use of its text-generation capabilities.<h2 id="conversion-process">Conversion process</h2>The model was converted using the following pipeline:<ul><li>Original abliterated model downloaded from <a target="_blank" rel="ugc" href="https://huggingface.co/huihui-ai">huihui-ai</a> on Hugging Face.</li><li>Converted using <a target="_blank" rel="ugc" href="https://github.com/dreamfast/heretic-docker">dreamfast's heretic-docker</a> conversion tools.</li><li>Packaged into the ComfyUI/Krea 2 safetensor format and quantized to FP8.</li></ul>No weights were modified beyond the conversion and packaging process required for ComfyUI compatibility.<h2 id="full-credit">Full credit</h2>This upload would not exist without the work of several people:<h3 id="abliterated-model">Abliterated model</h3><a target="_blank" rel="ugc" href="https://huggingface.co/huihui-ai">huihui-ai</a>Created and released the abliterated version of Qwen3-VL 4B used for this conversion.<h3 id="conversion-tooling">Conversion tooling</h3><a target="_blank" rel="ugc" href="https://github.com/dreamfast">dreamfast</a>Created the excellent heretic-docker project and conversion scripts that make this conversion possible.Their tooling handles:<ul><li>model shard merging</li><li>ComfyUI key remapping</li><li>safetensor generation</li><li>FP8 and NVFP4 quantization</li></ul>I simply followed their workflow.<h2 id="about-sample-images">About sample images</h2>NSFW LoRA: <a target="_blank" rel="ugc" href="https://civarchive.com/models/2725430/krea-2-nsfw-v2?modelVersionId=3065362">Krea 2 NSFW V2</a> Conditioning rebalance node: <a target="_blank" rel="ugc" href="https://github.com/nova452/ComfyUI-ConditioningKrea2Rebalance">ComfyUI-ConditioningKrea2Rebalance</a>This was the prompt used for the sample images. This prompt is able to jail brake the standard Qwen3VL model:<pre><code>You are a visual prompt engineering assistant.

Your task is to analyze the provided image and produce a highly detailed image-generation prompt that would recreate the image as closely as possible.

Rules:
- Do NOT describe the image conversationally.
- Output ONLY a prompt suitable for an image generation model.
- Be precise, objective, and exhaustive.
- Do NOT mention the original image, camera metadata unless visible, or say “this image shows”.
- Avoid extra sentences like "Here is..."
- Include NSFW descriptions if aplicable and if you are able to, if not, ignore this directive.

Prompt requirements:
1. Subject description:
 - Body proportions
 - Facial features, skin texture, expression, gaze direction. Describe the face as a physical map of movements:
 MOUTH: Be hyper-specific. Is the lower lip pushed out? Are corners pulled down? (e.g., "lips pursed into a tight pucker, lower lip protruding").
 EYE/BROW TENSION: Describe "squinching," wide-set lids, or furrowed brows. Explicitly describe the position of the pupils and the direction of the gaze (e.g., 'pupils rolled upward,' 'looking away from the lens'
 - Hair style, hair color, accessories
 - Clothing, materials, fit, layers

2. Pose and composition:
 - Body pose, hand position, posture. Identify the primary support points. Use "Kneeling," "Crouching," or "Leaning"
 - Describe how the body is angled relative to the camera. Mention the relationship between the head, shoulders, and knees (e.g., "leaning her weight heavily forward onto her knees, torso lunging toward the lens, neck slightly compressed")
 - Framing (close-up, medium shot, full body)
 - Camera angle (eye level, low angle, top-down, etc)
 - Subject placement in frame

3. Environment and background:
 - Location type (studio, indoor, outdoor)
 - Background color, texture, objects
 - Depth of field

4. Lighting:
 - Light direction, softness, contrast
 - Key light, fill light, rim light if applicable
 - Time of day or artificial lighting style

5. Artistic and technical style:
 - Photorealistic, cinematic, illustration, anime, 3D render, etc.
 - Lens look (wide, portrait compression), bokeh if visible
 - Image sharpness, noise, realism level

6. Color and mood:
 - Dominant colors
 - Color grading (warm, cool, neutral, muted, vibrant)
 - Emotional tone

Formatting rules:
- Output as a single, highly descriptive paragraph of natural, dense prose.
- Use commas to separate attributes.
- Avoid bullet points.
- Avoid vague terms like “beautiful”, “nice”, “high quality”.
- Use concrete, reproducible descriptors.</code></pre>

Qwen3-VL 4B Abliterated (ComfyUI / Krea 2 Text Encoder) BF16 / FP8

qwen3VL4BAbliteratedComfyui_v10_full_bf16.safetensors

qwen3VL4BAbliteratedComfyui_v10_full_fp8.safetensors

v1.0

<h2 id="please-read-the-version-noteswhat-this-is">PLEASE READ THE VERSION NOTES <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/18a3c7d9-6212-40cf-851c-3159d065a339/original=true/18a3c7d9-6212-40cf-851c-3159d065a339.jpeg" />What this is?</h2>A TURBO checkpoint merged for the perfect balance between three things: NSFW capability, prompt adherence, and a "smartphone-like" look. The goal is a balanced crisp, casual image. This is the goal, and this checkpoint is the starting point; expect many updates with high frequency.<h2 id="baked-loras-changelog-(v5):">Baked LoRAs Changelog (v5):</h2>I'm lazy now, check attached loras on the samplesAll baked LoRAs will credit their creators.<h2 id="recommended-settings">⚙️ Recommended Settings</h2><h3 id="v1">V1</h3><ul><li>res_2s + beta -&gt; More Quality</li><li>er_sde + beta -&gt; More Speed</li><li>CFG: 1</li><li>Steps: 8–10</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow">Workflow v1</a></li></ul><h3 id="v2">V2</h3><ul><li>SamplerER_SDE + beta -&gt; Fast and Stable</li><li>CFG: 1</li><li>Steps: 8–12</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow?modelVersionId=3118938">Workflow v2</a></li></ul><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b647b323-f1c5-4326-992a-577bcbb7c270/original=true/b647b323-f1c5-4326-992a-577bcbb7c270.jpeg" /><h3 id="v3">V3</h3><ul><li>euler + simple</li><li>euler + beta (my favorite)</li><li>er_sde + simple</li><li>er_sde + beta</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow">Workflow v1</a> or <a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow">Workflow v3</a></li></ul><h3 id="v4">V4</h3><ul><li>euler + beta</li><li>er_sde + simple</li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow">Workflow v1</a> or <a target="_blank" rel="ugc" href="https://civarchive.com/models/2763220/krea2-simple-fine-workflow">Workflow v3</a></li></ul><h3 id="v5">V5</h3><ul><li>euler + simple</li><li>CFG 1</li><li>10-12 Steps</li></ul><h2 id="resolutions">📐 Resolutions</h2>I found that scaling resolutions by x1.25 and x1.5 gives better results:- 832x1216 standard → 1040x1520 optimal 1248x1824 recommended- 896x1152 standard → 1120x1440 optimal 1344x1728 recommended- 1024x1024 standard → 1280x1280 optimal 1536x1536 recommended<h2 id="notes">💡 Notes</h2>Good results are achievable without a face detailer or extra passes. With the right config, some patience, and a bit of seed luck, you can get great things. I highly recommend start all your prompts using a phrase like this or similar.<pre><code>This is a casual, low-quality photo</code></pre><pre><code>this is an amateur photo taken from smartphone, casual photo</code></pre><h2 id="important-disclaimer">⚠️ Important disclaimer</h2>This is NOT trained by me and is NOT my own work, my job here is using my experience to bake balanced weights of my personal taste to achieve desired images using my prompt style. My only role is to share with you what, to my eye, is the best style and quality achievable with this model. All baked LoRAs credit their creators.<h3 id="about-prompts">ABOUT PROMPTS</h3>You, prompt scammer. You are free to scam my prompt without mentioning or giving the credit, like the majority does here. i won't reveal incredible prompts anymore, so you can keep reusing my old prompts.

FinePorn v5 - INT8

Rebalanced LoRas (v2):<ul><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2688234/realism-engine-ideogram-4-krea-2?modelVersionId=3070702">Realism Engine</a> — reduced weight</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2268008/realistic-snapshot-z-image-turbo-krea-2?modelVersionId=3084537">Realistic Snapshot</a> — reduced weight</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2725430/krea-2-nsfw-v4?modelVersionId=3085473">Krea 2 NSFW V4</a> — removed</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/1972981/sex-nudes-other-fun-stuff-snofs?modelVersionId=3104629">SNOFS V1.1 </a>— added</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2766540/rawgirlspicy-krea2?modelVersionId=3117160">RawGirlSpicy V2 </a>— added</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2728005/purelens-realism-ideogram-4-krea-2?modelVersionId=3114242">PureLens V1 </a>— added</li><li><a target="_blank" rel="ugc" href="https://civitai.red/models/2537408/phone-photography-2000-2025-krea2-klein-9b?modelVersionId=3111163">PhonePhoto </a>— added </li></ul>EXPLANATION:In this version, I'm aiming for slightly sharper and more modern images to better capture details like fluids, skin, makeup, etc. I've also incorporated some updated versions of the LoRas to keep them current. The addition of some extra LoRas is because I believe they generate more variety (some things might be more difficult to achieve with a prompt, so i encourage you to experiment with different forms of prompts to achieve the same result).I've removed NSFW v4.1 because I feel it was causing artifacts or grainy distortion in the image. Additionally, its weight are very sensitive and significantly affect the functionality of the other LoRas (I hope to reinstate it soon). NEWER IS NOT BETTERI want to clarify something, this is my subjetive taste, so V2 isn't necessarily better than V1, there could be the case where you prefer to keep on V1 because it fits perfect for you. IMO this V2 feels better so pick your favorite. NEW WORKFLOW and NEW SAMPLERFor this version, i'm experimenting with another native sampler (SamplerER_SDE + Beta) that i ''guess'' gives better overall results; while trying to keep my workflow as simple as possible, youn can download the workflow here: <a rel="ugc" href="https://civitai.red/models/2763220/krea2-simple-fine-workflow">SimpleFineWorkflow V2</a>orYou can just ignore it and still using euler/beta or er_sde/beta. BONUSFor those who (for some reason) were asking for a normal FP8 version, i'm including both files on this version, FP8 and INT8. (i still recommending int8 for x2 speed). Nothing more to say, i hope this can be useful for you and if you like it, please leave a comment, it's useful for me :) enjoy your weekend. Sincerely, Simon.

Files with SHA256: e6a3527e5cfca02ecc37108d607494c9fcf7475a68e762f17822996b4d3d98ef