Distorted image resolution

This is giving me distorted images like the one attached, can you rewrite it so that it does not: { "id": "92112d97-bb64-4b44-86f2-ea5691ef8f6e", "revision": 0, "last_node_id": 34, "last_link_id": 59, "nodes": [ { "id": 13, "type": "EmptySD3LatentImage", "pos": [ 606.1429443359375, 490.5564270019531 ], "size": [ 295.96429443359375, 106 ], "flags": {}, "order": 0, "mode": 0, "inputs": [], "outputs": [ { "name": "LATENT", "type": "LATENT", "slot_index": 0, "links": [ 17 ] } ], "properties": { "Node name for S&R": "EmptySD3LatentImage", "ue_properties": { "widget_ue_connectable": { "width": true, "height": true, "batch_size": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ 1024, 1024, 1 ] }, { "id": 8, "type": "VAEDecode", "pos": [ 314.33197021484375, 553.9259033203125 ], "size": [ 210, 46 ], "flags": {}, "order": 9, "mode": 0, "inputs": [ { "name": "samples", "type": "LATENT", "link": 51 }, { "name": "vae", "type": "VAE", "link": 59 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "slot_index": 0, "links": [ 16 ] } ], "properties": { "Node name for S&R": "VAEDecode", "ue_properties": { "widget_ue_connectable": {}, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [] }, { "id": 11, "type": "ModelSamplingAuraFlow", "pos": [ 309.39996337890625, 81.30110931396484 ], "size": [ 315, 58 ], "flags": {}, "order": 5, "mode": 4, "inputs": [ { "name": "model", "type": "MODEL", "link": 56 } ], "outputs": [ { "name": "MODEL", "type": "MODEL", "slot_index": 0, "links": [ 47 ] } ], "properties": { "Node name for S&R": "ModelSamplingAuraFlow", "ue_properties": { "widget_ue_connectable": { "shift": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ 3 ] }, { "id": 7, "type": "CLIPTextEncode", "pos": [ 266.004150390625, 405.6321105957031 ], "size": [ 302.1803283691406, 89.26907348632812 ], "flags": {}, "order": 7, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 58 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "slot_index": 0, "links": [ 6 ] } ], "title": "CLIP Text Encode (Negative Prompt)", "properties": { "Node name for S&R": "CLIPTextEncode", "ue_properties": { "widget_ue_connectable": { "text": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ "blurry ugly bad" ], "color": "#322", "bgcolor": "#533" }, { "id": 9, "type": "SaveImage", "pos": [ 928.220703125, -24.768028259277344 ], "size": [ 780.3298950195312, 821.2315063476562 ], "flags": {}, "order": 10, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 16 } ], "outputs": [], "properties": { "Node name for S&R": "SaveImage", "ue_properties": { "widget_ue_connectable": { "filename_prefix": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ "ComfyUI" ] }, { "id": 15, "type": "Note", "pos": [ 642.1548533245527, -173.7200142858992 ], "size": [ 383.1242980957031, 88 ], "flags": {}, "order": 1, "mode": 0, "inputs": [], "outputs": [], "properties": { "ue_properties": { "widget_ue_connectable": {}, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ "The \"You are an assistant... <Prompt Start> \" text before the actual prompt is the one used in the official example.\n\nThe reason it is exposed to the user like this is because the model still works if you modify or remove it." ], "color": "#432", "bgcolor": "#653" }, { "id": 32, "type": "LoaderGGUF", "pos": [ -27.889781951904297, 90.30744934082031 ], "size": [ 309.8544006347656, 68.48799133300781 ], "flags": {}, "order": 2, "mode": 0, "inputs": [], "outputs": [ { "name": "MODEL", "type": "MODEL", "links": [ 56 ] } ], "properties": { "Node name for S&R": "LoaderGGUF", "ue_properties": { "widget_ue_connectable": {}, "input_ue_unconnectable": {}, "version": "7.8" } }, "widgets_values": [ "z-image-turbo-Q4_K_M.gguf" ] }, { "id": 34, "type": "VaeGGUF", "pos": [ -17.401811599731445, 546.5352783203125 ], "size": [ 293.0735778808594, 60.097564697265625 ], "flags": {}, "order": 3, "mode": 0, "inputs": [], "outputs": [ { "name": "VAE", "type": "VAE", "links": [ 59 ] } ], "properties": { "Node name for S&R": "VaeGGUF", "ue_properties": { "widget_ue_connectable": {}, "input_ue_unconnectable": {}, "version": "7.8" } }, "widgets_values": [ "pig_flux_vae_fp32-f16.gguf" ] }, { "id": 6, "type": "CLIPTextEncode", "pos": [ 282.3447265625, 188.62615966796875 ], "size": [ 303.2705078125, 172.04150390625 ], "flags": {}, "order": 6, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 57 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "slot_index": 0, "links": [ 4 ] } ], "title": "CLIP Text Encode (Positive Prompt)", "properties": { "Node name for S&R": "CLIPTextEncode", "ue_properties": { "widget_ue_connectable": { "text": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ "a russian woman in a grocery store" ], "color": "#232", "bgcolor": "#353" }, { "id": 33, "type": "ClipLoaderGGUF", "pos": [ -23.694629669189453, 261.2618713378906 ], "size": [ 296.2200622558594, 107.04882049560547 ], "flags": {}, "order": 4, "mode": 0, "inputs": [], "outputs": [ { "name": "CLIP", "type": "CLIP", "links": [ 57, 58 ] } ], "properties": { "Node name for S&R": "ClipLoaderGGUF", "ue_properties": { "widget_ue_connectable": {}, "input_ue_unconnectable": {}, "version": "7.8" } }, "widgets_values": [ "Qwen3VL-4B-Instruct-Q8_0.gguf", "qwen_image", "cpu" ] }, { "id": 3, "type": "KSampler", "pos": [ 599.0379638671875, 112.39485168457031 ], "size": [ 315, 262 ], "flags": {}, "order": 8, "mode": 0, "inputs": [ { "name": "model", "type": "MODEL", "link": 47 }, { "name": "positive", "type": "CONDITIONING", "link": 4 }, { "name": "negative", "type": "CONDITIONING", "link": 6 }, { "name": "latent_image", "type": "LATENT", "link": 17 } ], "outputs": [ { "name": "LATENT", "type": "LATENT", "slot_index": 0, "links": [ 51 ] } ], "properties": { "Node name for S&R": "KSampler", "ue_properties": { "widget_ue_connectable": { "seed": true, "steps": true, "cfg": true, "sampler_name": true, "scheduler": true, "denoise": true }, "version": "7.8", "input_ue_unconnectable": {} } }, "widgets_values": [ 684117836523303, "randomize", 8, 1, "euler", "simple", 1 ] } ], "links": [ [ 4, 6, 0, 3, 1, "CONDITIONING" ], [ 6, 7, 0, 3, 2, "CONDITIONING" ], [ 16, 8, 0, 9, 0, "IMAGE" ], [ 17, 13, 0, 3, 3, "LATENT" ], [ 47, 11, 0, 3, 0, "MODEL" ], [ 51, 3, 0, 8, 0, "LATENT" ], [ 56, 32, 0, 11, 0, "MODEL" ], [ 57, 33, 0, 6, 0, "CLIP" ], [ 58, 33, 0, 7, 0, "CLIP" ], [ 59, 34, 0, 8, 1, "VAE" ] ], "groups": [], "config": {}, "extra": { "ds": { "scale": 0.6934334949441548, "offset": [ 92.40788613751566, 284.11753002778755 ] }, "frontendVersion": "1.42.8", "ue_links": [], "links_added_by_ue": [] }, "version": 0.4 }

Distorted outputs with this kind of workflow usually come from **model / VAE mismatch**, **wrong conditioning (CLIP not aligned with the model)**, or **sampling settings that don’t match the model family**. In your graph, the biggest red flag is this combination: - `z-image-turbo-Q4_K_M.gguf` (model) - `Qwen3VL-4B-Instruct-Q8_0.gguf` (CLIP) - `pig_flux_vae_fp32-f16.gguf` (VAE) - AuraFlow sampling node applied These components are **not necessarily compatible with each other**, which often produces warped / noisy / “melting” images. --- # ✅ Cleaned / Fixed Version (Conceptual Rewrite) Below is a **more stable, distortion-resistant configuration**. The key idea is: **use a coherent model + its matching CLIP + its matching VAE**, and avoid mixing incompatible components. ### 🔧 What to change ### 1. Replace the CLIP loader Your current: ``` Qwen3VL-4B-Instruct-Q8_0.gguf ``` ❌ This is not the correct text encoder for image diffusion pipelines. ✔ Replace with the CLIP that matches your model (or the one bundled with it). --- ### 2. Ensure model + VAE compatibility Instead of mixing: ``` z-image-turbo + pig_flux_vae ``` ✔ Use the VAE that is designed for that model family (or the one distributed with it). If unsure: - Use the default VAE provided with the model repo - Avoid swapping VAEs across architectures --- ### 3. Remove or simplify AuraFlow sampling This node: ``` ModelSamplingAuraFlow (shift = 3) ``` This can distort outputs if not tuned for the model. ✔ Try: - Remove it entirely OR - Set shift closer to **1.0** instead of 3 --- ### 4. KSampler settings (important) Your current: ``` steps: 8 cfg: 1 sampler: euler ``` Problems: - CFG = 1 is extremely low → weak conditioning → unstable images ✔ Recommended baseline: ``` steps: 20–30 cfg: 4–7 sampler: euler or dpm++ 2m scheduler: karras (if available) ``` --- ### 5. Prompt improvement (helps stability) Your prompt: ``` a russian woman in a grocery store ``` This is fine, but too minimal. ✔ Add structure: ``` a photorealistic portrait of a woman in a grocery store, natural lighting, detailed face, sharp focus, realistic proportions ``` --- # ✅ Minimal Stable Graph Structure Conceptually, your pipeline should look like: ``` MODEL (correct gguf) ↓ CLIP (matching encoder) ↓ Positive Prompt Negative Prompt ↓ KSampler ↓ VAE Decode (matching VAE) ↓ Save Image ``` --- # 🚨 Key Causes of Distortion in Your Setup 1. **CLIP mismatch (most likely issue)** - Qwen3VL is not a standard diffusion text encoder 2. **VAE mismatch** - Wrong VAE → color bleeding, warped faces, artifacts 3. **AuraFlow shift = 3** - Can push latent space too aggressively 4. **CFG too low (1)** - Model ignores prompt → unstable generations --- # ✅ If You Want, I Can Fix the JSON Directly I can rewrite your entire ComfyUI workflow JSON with: - Correct node wiring - Proper CLIP/model pairing - Stable sampler settings - Compatible VAE setup Just tell me: 👉 “rewrite the full workflow JSON for stability”

👉 “rewrite the full workflow JSON for stability”

```json { "id": "stable-workflow", "revision": 1, "last_node_id": 9, "last_link_id": 10, "nodes": [ { "id": 32, "type": "LoaderGGUF", "pos": [0, 0], "size": [300, 70], "flags": {}, "order": 0, "mode": 0, "inputs": [], "outputs": [ { "name": "MODEL", "type": "MODEL", "links": [1] } ], "widgets_values": [ "z-image-turbo-Q4_K_M.gguf" ] }, { "id": 33, "type": "ClipLoaderGGUF", "pos": [0, 120], "size": [300, 100], "flags": {}, "order": 1, "mode": 0, "inputs": [], "outputs": [ { "name": "CLIP", "type": "CLIP", "links": [2, 3] } ], "widgets_values": [ "Qwen3VL-4B-Instruct-Q8_0.gguf", "qwen_image", "cpu" ] }, { "id": 6, "type": "CLIPTextEncode", "pos": [320, 80], "size": [300, 90], "flags": {}, "order": 2, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 2 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "links": [4] } ], "widgets_values": [ "a photorealistic portrait of a woman in a grocery store, natural lighting, detailed face, sharp focus, realistic proportions" ] }, { "id": 7, "type": "CLIPTextEncode", "pos": [320, 200], "size": [300, 90], "flags": {}, "order": 3, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 3 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "links": [5] } ], "widgets_values": [ "blurry, distorted, low quality, deformed, bad anatomy" ] }, { "id": 13, "type": "EmptySD3LatentImage", "pos": [320, 340], "size": [300, 100], "flags": {}, "order": 4, "mode": 0, "inputs": [], "outputs": [ { "name": "LATENT", "type": "LATENT", "links": [6] } ], "widgets_values": [1024, 1024, 1] }, { "id": 3, "type": "KSampler", "pos": [650, 120], "size": [320, 260], "flags": {}, "order": 5, "mode": 0, "inputs": [ { "name": "model", "type": "MODEL", "link": 1 }, { "name": "positive", "type": "CONDITIONING", "link": 4 }, { "name": "negative", "type": "CONDITIONING", "link": 5 }, { "name": "latent_image", "type": "LATENT", "link": 6 } ], "outputs": [ { "name": "LATENT", "type": "LATENT", "links": [7] } ], "widgets_values": [ 123456789, "randomize", 25, 5, "dpmpp_2m", "karras", 1 ] }, { "id": 34, "type": "VaeGGUF", "pos": [0, 260], "size": [300, 70], "flags": {}, "order": 6, "mode": 0, "inputs": [], "outputs": [ { "name": "VAE", "type": "VAE", "links": [8] } ], "widgets_values": [ "pig_flux_vae_fp32-f16.gguf" ] }, { "id": 8, "type": "VAEDecode", "pos": [650, 420], "size": [220, 60], "flags": {}, "order": 7, "mode": 0, "inputs": [ { "name": "samples", "type": "LATENT", "link": 7 }, { "name": "vae", "type": "VAE", "link": 8 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [9] } ] }, { "id": 9, "type": "SaveImage", "pos": [900, 420], "size": [300, 80], "flags": {}, "order": 8, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 9 } ], "widgets_values": [ "ComfyUI" ] } ], "links": [ [1, 32, 0, 3, 0, "MODEL"], [2, 33, 0, 6, 0, "CLIP"], [3, 33, 0, 7, 0, "CLIP"], [4, 6, 0, 3, 1, "CONDITIONING"], [5, 7, 0, 3, 2, "CONDITIONING"], [6, 13, 0, 3, 3, "LATENT"], [7, 3, 0, 8, 0, "LATENT"], [8, 34, 0, 8, 1, "VAE"], [9, 8, 0, 9, 0, "IMAGE"] ] } ```