ostris/flex-redux 🔢📝🖼️❓ → 🖼️
About
A redux adapter trained from scratch on Flex.1-alpha, that also works with FLUX.1-dev
Example Output
Prompt:
"a photo of a woman"
Output
Performance Metrics
23.23s
Prediction Time
186.97s
Total Time
All Input Parameters
{
"width": 1024,
"height": 1024,
"prompt": "a photo of a woman",
"guidance": 3,
"redux_image": "https://replicate.delivery/pbxt/MlZs8vm1bTgi0g4ozCqINiulCLkb3w0gln1TargqaiTz1L4Q/0_1.webp",
"aspect_ratio": "1:1",
"output_format": "webp",
"output_quality": 95,
"redux_strength": 1,
"clip_vision_crop": "center",
"redux_strength_type": "multiply"
}
Input Parameters
- seed
- Set a seed for reproducibility. Random by default.
- width
- Width of the output image. Ignored if aspect ratio is not None.
- height
- Height of the output image. Ignored if aspect ratio is not None.
- prompt
- guidance
- Guidance scale for the redux image.
- redux_image (required)
- Input image to condition your output on.
- aspect_ratio
- Aspect ratio of the output image.
- output_format
- Format of the output images
- output_quality
- Quality of the output images, from 0 to 100. 100 is best quality, 0 is lowest quality.
- redux_strength
- Strength of the redux image. 0 = no redux, 1 = full redux.
- clip_vision_crop
- Whether to crop the redux image for clip vision. Crops to a square if center, otherwise no cropping. Images can appear distorted if no cropping is used.
- redux_strength_type
- Type of strength for the redux image.
Output Schema
Output
Example Execution Logs
Random seed set to: 2800627173
Checking inputs
✅ /tmp/inputs/image.webp
====================================
Checking weights
✅ flex1_redux_siglip2_512.safetensors exists in ComfyUI/models/style_models
✅ siglip2_so400m_patch16_512.safetensors exists in ComfyUI/models/clip_vision
✅ Flex.1-alpha.safetensors exists in ComfyUI/models/checkpoints
====================================
Running workflow
[ComfyUI] got prompt
Executing node 44, title: Load Checkpoint, class type: CheckpointLoaderSimple
[ComfyUI] model weight dtype torch.bfloat16, manual cast: None
[ComfyUI] model_type FLUX
[ComfyUI] Using pytorch attention in VAE
[ComfyUI] Using pytorch attention in VAE
[ComfyUI] VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[ComfyUI] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
Executing node 27, title: EmptySD3LatentImage, class type: EmptySD3LatentImage
Executing node 30, title: ModelSamplingFlux, class type: ModelSamplingFlux
Executing node 17, title: BasicScheduler, class type: BasicScheduler
Executing node 16, title: KSamplerSelect, class type: KSamplerSelect
Executing node 40, title: Load Image, class type: LoadImage
Executing node 46, title: Load Advanced Vision Model, class type: AdvancedVisionLoader
[ComfyUI] Requested to load CLIPVisionModelProjection
Executing node 39, title: CLIP Vision Encode, class type: CLIPVisionEncode
[ComfyUI] loaded completely 43939.05 788.7587585449219 True
Executing node 42, title: Load Style Model, class type: StyleModelLoader
Executing node 6, title: CLIP Text Encode (Positive Prompt), class type: CLIPTextEncode
[ComfyUI] Requested to load FluxClipModel_
[ComfyUI] loaded completely 43042.16624145508 4777.53759765625 True
Executing node 26, title: FluxGuidance, class type: FluxGuidance
Executing node 41, title: Apply Style Model, class type: StyleModelApply
Executing node 22, title: BasicGuider, class type: BasicGuider
Executing node 25, title: RandomNoise, class type: RandomNoise
Executing node 13, title: SamplerCustomAdvanced, class type: SamplerCustomAdvanced
[ComfyUI] Requested to load Flux
[ComfyUI] loaded completely 38164.32463616943 15570.190551757812 True
[ComfyUI] Advanced Vision Model: clip_vision_siglip2_so400m_512 detected
[ComfyUI]
[ComfyUI] 0%| | 0/30 [00:00<?, ?it/s]
[ComfyUI] 3%|▎ | 1/30 [00:00<00:16, 1.74it/s]
[ComfyUI] 7%|▋ | 2/30 [00:01<00:15, 1.85it/s]
[ComfyUI] 10%|█ | 3/30 [00:01<00:14, 1.89it/s]
[ComfyUI] 13%|█▎ | 4/30 [00:02<00:13, 1.91it/s]
[ComfyUI] 17%|█▋ | 5/30 [00:02<00:13, 1.92it/s]
[ComfyUI] 20%|██ | 6/30 [00:03<00:12, 1.92it/s]
[ComfyUI] 23%|██▎ | 7/30 [00:03<00:11, 1.93it/s]
[ComfyUI] 27%|██▋ | 8/30 [00:04<00:11, 1.93it/s]
[ComfyUI] 30%|███ | 9/30 [00:04<00:10, 1.93it/s]
[ComfyUI] 33%|███▎ | 10/30 [00:05<00:10, 1.93it/s]
[ComfyUI] 37%|███▋ | 11/30 [00:05<00:09, 1.93it/s]
[ComfyUI] 40%|████ | 12/30 [00:06<00:09, 1.93it/s]
[ComfyUI] 43%|████▎ | 13/30 [00:06<00:08, 1.93it/s]
[ComfyUI] 47%|████▋ | 14/30 [00:07<00:08, 1.93it/s]
[ComfyUI] 50%|█████ | 15/30 [00:07<00:07, 1.93it/s]
[ComfyUI] 53%|█████▎ | 16/30 [00:08<00:07, 1.93it/s]
[ComfyUI] 57%|█████▋ | 17/30 [00:08<00:06, 1.93it/s]
[ComfyUI] 60%|██████ | 18/30 [00:09<00:06, 1.93it/s]
[ComfyUI] 63%|██████▎ | 19/30 [00:09<00:05, 1.93it/s]
[ComfyUI] 67%|██████▋ | 20/30 [00:10<00:05, 1.93it/s]
[ComfyUI] 70%|███████ | 21/30 [00:10<00:04, 1.93it/s]
[ComfyUI] 73%|███████▎ | 22/30 [00:11<00:04, 1.93it/s]
[ComfyUI] 77%|███████▋ | 23/30 [00:11<00:03, 1.92it/s]
[ComfyUI] 80%|████████ | 24/30 [00:12<00:03, 1.92it/s]
[ComfyUI] 83%|████████▎ | 25/30 [00:13<00:02, 1.92it/s]
[ComfyUI] 87%|████████▋ | 26/30 [00:13<00:02, 1.92it/s]
[ComfyUI] 90%|█████████ | 27/30 [00:14<00:01, 1.92it/s]
[ComfyUI] 93%|█████████▎| 28/30 [00:14<00:01, 1.92it/s]
[ComfyUI] 97%|█████████▋| 29/30 [00:15<00:00, 1.92it/s]
[ComfyUI] 100%|██████████| 30/30 [00:15<00:00, 1.92it/s]
[ComfyUI] 100%|██████████| 30/30 [00:15<00:00, 1.92it/s]
[ComfyUI] Requested to load AutoencodingEngine
Executing node 8, title: VAE Decode, class type: VAEDecode
[ComfyUI] loaded completely 19149.599937438965 159.87335777282715 True
Executing node 49, title: Save Image, class type: SaveImage
[ComfyUI] Prompt executed in 22.79 seconds
outputs: {'49': {'images': [{'filename': 'R8_Flex_Redux_00001_.png', 'subfolder': '', 'type': 'output'}]}}
====================================
R8_Flex_Redux_00001_.png
Version Details
- Version ID
ed37ab87cb1dfe4cc5c42c078298765aadc7cf476468acc1c8df5c9996fefccd- Version Created
- April 2, 2025