{ "id": "5140b129-e927-49a6-816e-668c69d43008", "revision": 0, "last_node_id": 86, "last_link_id": 152, "nodes": [ { "id": 1, "type": "UNETLoader", "pos": [ -440, -410 ], "size": [ 500, 90 ], "flags": {}, "order": 0, "mode": 0, "inputs": [], "outputs": [ { "name": "MODEL", "type": "MODEL", "slot_index": 0, "links": [ 91 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "UNETLoader", "models": [ { "name": "moss_image_sft_int8_convrot.safetensors", "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors", "directory": "diffusion_models" } ] }, "widgets_values": [ "moss_image_sft_int8_convrot.safetensors", "default" ], "widgets_values_named": { "unet_name": "moss_image_sft_int8_convrot.safetensors", "weight_dtype": "default" } }, { "id": 2, "type": "SkipLayerGuidanceDiTSimple", "pos": [ 620, -460 ], "size": [ 400, 130 ], "flags": {}, "order": 19, "mode": 0, "showAdvanced": true, "inputs": [ { "name": "model", "type": "MODEL", "link": 69 } ], "outputs": [ { "name": "MODEL", "type": "MODEL", "slot_index": 0, "links": [ 35, 36 ] } ], "title": "SkipLayerGuidanceDiTSimple (uncond skips block 22)", "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "SkipLayerGuidanceDiTSimple" }, "widgets_values": [ "", "24", 0, 1 ], "widgets_values_named": { "double_layers": "", "single_layers": "24", "start_percent": 0, "end_percent": 1 } }, { "id": 3, "type": "CLIPLoader", "pos": [ -440, -50 ], "size": [ 500, 110 ], "flags": {}, "order": 1, "mode": 0, "inputs": [], "outputs": [ { "name": "CLIP", "type": "CLIP", "slot_index": 0, "links": [ 3, 4 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "CLIPLoader", "models": [ { "name": "qwen3.5_9b_int8_convrot.safetensors", "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors", "directory": "text_encoders" } ] }, "widgets_values": [ "qwen3.5_9b_w4a8_gptq.safetensors", "moss_image", "default" ], "widgets_values_named": { "clip_name": "qwen3.5_9b_w4a8_gptq.safetensors", "type": "moss_image", "device": "default" } }, { "id": 4, "type": "VAELoader", "pos": [ -440, 130 ], "size": [ 500, 60 ], "flags": {}, "order": 2, "mode": 0, "inputs": [], "outputs": [ { "name": "VAE", "type": "VAE", "slot_index": 0, "links": [ 9, 105 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "VAELoader", "models": [ { "name": "flux2-vae_small.safetensors", "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors", "directory": "vae" } ] }, "widgets_values": [ "flux2-vae_small.safetensors" ], "widgets_values_named": { "vae_name": "flux2-vae_small.safetensors" } }, { "id": 5, "type": "CLIPTextEncode", "pos": [ 150, -60 ], "size": [ 470, 240 ], "flags": {}, "order": 10, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 3 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "slot_index": 0, "links": [ 99 ] } ], "title": "Positive Prompt", "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "CLIPTextEncode" }, "widgets_values": [ "Full-body photograph of a knight in polished steel plate armour with a long red cloak, standing in a sunlit stone castle courtyard, banners on the walls behind him." ], "widgets_values_named": { "text": "Full-body photograph of a knight in polished steel plate armour with a long red cloak, standing in a sunlit stone castle courtyard, banners on the walls behind him." }, "color": "#232", "bgcolor": "#353" }, { "id": 6, "type": "CLIPTextEncode", "pos": [ 150, 250 ], "size": [ 480, 110 ], "flags": {}, "order": 11, "mode": 0, "inputs": [ { "name": "clip", "type": "CLIP", "link": 4 } ], "outputs": [ { "name": "CONDITIONING", "type": "CONDITIONING", "slot_index": 0, "links": [ 100 ] } ], "title": "Negative Prompt", "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "CLIPTextEncode" }, "widgets_values": [ "nsfw, worst quality, bad anatomy, watermark, jpeg artifact" ], "widgets_values_named": { "text": "nsfw, worst quality, bad anatomy, watermark, jpeg artifact" }, "color": "#223", "bgcolor": "#335" }, { "id": 9, "type": "VAEDecode", "pos": [ 1910, -520 ], "size": [ 210, 50 ], "flags": {}, "order": 23, "mode": 0, "inputs": [ { "name": "samples", "type": "LATENT", "link": 25 }, { "name": "vae", "type": "VAE", "link": 9 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "slot_index": 0, "links": [ 119, 151 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "VAEDecode" }, "color": "#322", "bgcolor": "#533" }, { "id": 16, "type": "TemporalScoreRescaling", "pos": [ 160, -470 ], "size": [ 400, 90 ], "flags": {}, "order": 13, "mode": 0, "showAdvanced": true, "inputs": [ { "name": "model", "type": "MODEL", "link": 117 } ], "outputs": [ { "name": "patched_model", "type": "MODEL", "links": [ 59 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "TemporalScoreRescaling" }, "widgets_values": [ 0.95, 1 ], "widgets_values_named": { "tsr_k": 0.95, "tsr_sigma": 1 } }, { "id": 20, "type": "SamplerCustom", "pos": [ 1500, -360 ], "size": [ 280, 500 ], "flags": {}, "order": 22, "mode": 0, "inputs": [ { "name": "model", "type": "MODEL", "link": 35 }, { "name": "positive", "type": "CONDITIONING", "link": 101 }, { "name": "negative", "type": "CONDITIONING", "link": 102 }, { "name": "sampler", "type": "SAMPLER", "link": 94 }, { "name": "sigmas", "type": "SIGMAS", "link": 27 }, { "name": "latent_image", "type": "LATENT", "link": 106 } ], "outputs": [ { "name": "output", "type": "LATENT", "links": [ 25 ] }, { "name": "denoised_output", "type": "LATENT", "links": null } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "SamplerCustom" }, "widgets_values": [ true, 42, "fixed", 3 ], "widgets_values_named": { "add_noise": true, "noise_seed": 42, "control_after_generate": "randomize", "cfg": 3 } }, { "id": 21, "type": "SamplerER_SDE", "pos": [ 1160, -310 ], "size": [ 270, 130 ], "flags": {}, "order": 3, "mode": 0, "showAdvanced": true, "inputs": [], "outputs": [ { "name": "SAMPLER", "type": "SAMPLER", "links": [ 94 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "SamplerER_SDE" }, "widgets_values": [ "ER-SDE", 3, 1, 1.05 ], "widgets_values_named": { "solver_type": "ER-SDE", "max_stage": 3, "eta": 1, "s_noise": 1.05 } }, { "id": 22, "type": "BasicScheduler", "pos": [ 1160, -90 ], "size": [ 270, 110 ], "flags": {}, "order": 21, "mode": 0, "inputs": [ { "name": "model", "type": "MODEL", "link": 36 } ], "outputs": [ { "name": "SIGMAS", "type": "SIGMAS", "links": [ 27 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "BasicScheduler" }, "widgets_values": [ "simple", 25, 1 ], "widgets_values_named": { "scheduler": "simple", "steps": 25, "denoise": 1 } }, { "id": 36, "type": "ModelSamplingAuraFlow", "pos": [ 160, -300 ], "size": [ 400, 90 ], "flags": {}, "order": 16, "mode": 0, "inputs": [ { "name": "model", "type": "MODEL", "link": 59 } ], "outputs": [ { "name": "MODEL", "type": "MODEL", "links": [ 69 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "ModelSamplingAuraFlow" }, "widgets_values": [ 5, "flow" ], "widgets_values_named": { "shift": 5, "sampling": "flow" } }, { "id": 42, "type": "PreviewImage", "pos": [ 1890, -380 ], "size": [ 890, 780 ], "flags": {}, "order": 24, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 119 } ], "outputs": [ { "name": "images", "type": "IMAGE", "links": [ 149 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "PreviewImage" } }, { "id": 49, "type": "LoraLoaderModelOnly", "pos": [ -440, -240 ], "size": [ 500, 120 ], "flags": {}, "order": 9, "mode": 4, "inputs": [ { "name": "model", "type": "MODEL", "link": 91 } ], "outputs": [ { "name": "MODEL", "type": "MODEL", "links": [ 117 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "LoraLoaderModelOnly", "models": [ { "name": "moss_image_color_depth_control.safetensors", "url": "https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors", "directory": "loras" } ] }, "widgets_values": [ "moss_image_color_depth_control.safetensors", 1 ], "widgets_values_named": { "lora_name": "moss_image_color_depth_control.safetensors", "strength_model": 1 } }, { "id": 53, "type": "MOSSImageControl", "pos": [ 770, -50 ], "size": [ 270, 190 ], "flags": {}, "order": 20, "mode": 0, "inputs": [ { "name": "positive", "type": "CONDITIONING", "link": 99 }, { "name": "negative", "type": "CONDITIONING", "link": 100 }, { "name": "vae", "type": "VAE", "link": 105 }, { "name": "color_map", "shape": 7, "type": "IMAGE", "link": null }, { "name": "depth_map", "shape": 7, "type": "IMAGE", "link": 148 } ], "outputs": [ { "name": "positive", "type": "CONDITIONING", "links": [ 101 ] }, { "name": "negative", "type": "CONDITIONING", "links": [ 102 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "MOSSImageControl" }, "widgets_values": [ 1, 0, 0.6 ], "widgets_values_named": { "strength": 1, "start_percent": 0, "end_percent": 0.6 } }, { "id": 54, "type": "EmptyLatentImage", "pos": [ 1510, 650 ], "size": [ 270, 110 ], "flags": {}, "order": 18, "mode": 0, "inputs": [ { "name": "width", "type": "INT", "widget": { "name": "width" }, "link": 125 }, { "name": "height", "type": "INT", "widget": { "name": "height" }, "link": 126 } ], "outputs": [ { "name": "LATENT", "type": "LATENT", "links": [ 106 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "EmptyLatentImage" }, "widgets_values": [ 1024, 1024, 1 ], "widgets_values_named": { "width": 1024, "height": 1024, "batch_size": 1 } }, { "id": 55, "type": "LoadImage", "pos": [ -400, 630 ], "size": [ 290, 110 ], "flags": {}, "order": 4, "mode": 0, "inputs": [], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 138 ] }, { "name": "MASK", "type": "MASK", "links": null } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "LoadImage" }, "widgets_values": [ "moss_cast_marcus.png", "image" ], "widgets_values_named": { "image": "moss_cast_marcus.png", "upload": "image" }, "color": "#322", "bgcolor": "#533" }, { "id": 61, "type": "GetImageSize", "pos": [ 1200, 680 ], "size": [ 140, 70 ], "flags": {}, "order": 15, "mode": 0, "inputs": [ { "name": "image", "type": "IMAGE", "link": 141 } ], "outputs": [ { "name": "width", "type": "INT", "links": [ 125 ] }, { "name": "height", "type": "INT", "links": [ 126 ] }, { "name": "batch_size", "type": "INT", "links": null } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "GetImageSize" } }, { "id": 76, "type": "bd7417be-5473-468c-99f1-78759bff8966", "pos": [ 400, 640 ], "size": [ 350, 190 ], "flags": {}, "order": 14, "mode": 0, "inputs": [ { "name": "image", "type": "IMAGE", "link": 140 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 147 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "previewExposures": [] }, "widgets_values": [ 9, 4, "depth", "moge_3_vitg_fp16.safetensors" ], "widgets_values_named": { "resolution_level": 9, "batch_size": 4, "output": "depth", "model_name": "moge_3_vitg_fp16.safetensors" } }, { "id": 77, "type": "ImageScaleToTotalPixels", "pos": [ 40, 630 ], "size": [ 280, 110 ], "flags": {}, "order": 12, "mode": 0, "showAdvanced": true, "inputs": [ { "name": "image", "type": "IMAGE", "link": 138 } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": [ 140, 141 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "ImageScaleToTotalPixels" }, "widgets_values": [ "lanczos", 1, 16 ], "widgets_values_named": { "upscale_method": "lanczos", "megapixels": 1, "resolution_steps": 16 } }, { "id": 79, "type": "PreviewImage", "pos": [ 810, 640 ], "size": [ 140, 30 ], "flags": {}, "order": 17, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 147 } ], "outputs": [ { "name": "images", "type": "IMAGE", "links": [ 148, 150 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "PreviewImage" } }, { "id": 80, "type": "ImageCompare", "pos": [ 2810, -370 ], "size": [ 930, 780 ], "flags": {}, "order": 26, "mode": 0, "inputs": [ { "name": "image_a", "shape": 7, "type": "IMAGE", "link": 150 }, { "name": "image_b", "shape": 7, "type": "IMAGE", "link": 149 } ], "outputs": [], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "ImageCompare" }, "widgets_values": [], "widgets_values_named": {} }, { "id": 81, "type": "Painter", "pos": [ -430, 1120 ], "size": [ 450, 550 ], "flags": {}, "order": 5, "mode": 0, "inputs": [ { "name": "image", "shape": 7, "type": "IMAGE", "link": null } ], "outputs": [ { "name": "IMAGE", "type": "IMAGE", "links": null }, { "name": "MASK", "type": "MASK", "links": null } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "Painter" }, "widgets_values": [ "", 512, 512, "#000000" ], "widgets_values_named": { "mask": "", "width": 512, "height": 512, "bg_color": "#000000" } }, { "id": 83, "type": "MarkdownNote", "pos": [ -1230, -520 ], "size": [ 710, 890 ], "flags": {}, "order": 6, "mode": 0, "inputs": [], "outputs": [], "title": "Note: About", "properties": {}, "widgets_values": [ "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n" ], "widgets_values_named": { "text": "## MOSS-Image: Color & Depth Control from a Photo\n\nDrop in a photo and the graph estimates its geometry, then generates a new image that keeps the same pose and framing. No map to prepare by hand.\n\n**How it works**\n\n1. **Load Image** (node 55) - the character reference. It is scaled to 1.0 MP, 16 px aligned (node 77), and drives the output size through **Get Image Size** (node 61) -> **Empty Latent** (node 54).\n2. **Geometry Estimation (MoGe 3)** subgraph - `Load MoGe Model` (`moge_3_vitg_fp16`) -> `Run MoGe Inference` -> `Render MoGe Geometry`. Its output is the `depth_map` of **MOSS-Image Color/Depth Control** (node 53).\n3. **Positive Prompt** (node 5) describes the character and the scene. Geometry comes from the render, everything else - identity, wardrobe, material, light, style - comes from the text.\n4. **MOSS-Image Color/Depth Control** (node 53) injects the map into the image tokens; the LoRA loaded just before it (`moss_image_color_depth_control.safetensors`, strength 1.0) is what makes the model read it. Without the LoRA the node is a no-op.\n5. **Save / Compare**: `VAE Decode` -> preview, plus **Compare Images** (node 80) holding the generated frame next to the rendered map.\n\n**Settings**\n\n`strength` 1.0 is the trained strength; `start_percent` / `end_percent` pick the part of the run that is controlled (0 to 0.6 here: the layout is fixed, the last steps render freely). Sampling follows the MOSS-Image control test graph: 25 steps, CFG 3, `er_sde` / `simple`, shift 5, skip layer guidance on layer 24, temporal score rescaling 0.95.\n\n**Useful knobs**\n\n- `Render MoGe Geometry` in the subgraph is set to `mask`: only the silhouette is used. Switch it to `depth` (`normalization = min_max`) when you want the face profile, shoulders and background falloff to be controlled too - the pose gets tighter, the prompt has less room.\n- `end_percent` 1.0 pins geometry to the last step; 0.5 frees the surface and lighting. For a mask-derived map 0.6 is a good middle.\n- The reference has **pointed elf ears**, and the geometry keeps them. Either keep that in the prompt (`pointed elf ears`) or hide them (`ears covered by hair`).\n- **Painter** (node 81) is in the graph for painting a map by hand: white near / black far for depth, one flat colour per region for colour. Connect its `IMAGE` to `depth_map` or `color_map` to override the estimate.\n- `color_map` is optional and currently unconnected. One flat colour per region, colours only act as region IDs, black means \"unspecified\"; it must match the output size (1024 x 1024 here).\n" }, "color": "#222", "bgcolor": "#000" }, { "id": 84, "type": "MarkdownNote", "pos": [ -1970, -520 ], "size": [ 520, 1250 ], "flags": {}, "order": 7, "mode": 0, "inputs": [], "outputs": [], "title": "Note: Model Links", "properties": {}, "widgets_values": [ "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\nšŸ“‚ ComfyUI/\nā”œā”€ā”€ šŸ“‚ models/\nā”‚ ā”œā”€ā”€ šŸ“‚ diffusion_models/\nā”‚ ā”‚ ā””ā”€ā”€ moss_image_sft_int8_convrot.safetensors 5.41 GB\nā”‚ ā”œā”€ā”€ šŸ“‚ text_encoders/\nā”‚ ā”‚ ā””ā”€ā”€ qwen3.5_9b_int8_convrot.safetensors 9.75 GB\nā”‚ ā”œā”€ā”€ šŸ“‚ vae/\nā”‚ ā”‚ ā””ā”€ā”€ flux2-vae_small.safetensors 249 MB\nā”‚ ā”œā”€ā”€ šŸ“‚ loras/\nā”‚ ā”‚ ā””ā”€ā”€ moss_image_color_depth_control.safetensors 313 MB\nā”‚ ā””ā”€ā”€ šŸ“‚ geometry_estimation/\nā”‚ ā””ā”€ā”€ moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n" ], "widgets_values_named": { "text": "## Model Links\n\n- [Hugging Face: Comfy-Org/MOSS-Image](https://huggingface.co/Comfy-Org/MOSS-Image)\n- [Hugging Face: Comfy-Org/MoGe](https://huggingface.co/Comfy-Org/MoGe) (geometry estimation)\n- [Upstream control adapter: darask0/MOSS-Image-color-depth-control](https://huggingface.co/darask0/MOSS-Image-color-depth-control)\n\n**diffusion_models**\n\n- [moss_image_sft_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_int8_convrot.safetensors) - 5.41 GB, the checkpoint this graph loads\n- [moss_image_sft_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_sft_bf16.safetensors) - 10.16 GB, higher precision, slower\n- [moss_image_base_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/diffusion_models/moss_image_base_int8_convrot.safetensors) - 5.41 GB, the base checkpoint runs in the same graph\n\n**text_encoders**\n\n- [qwen3.5_9b_int8_convrot.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_int8_convrot.safetensors) - 9.75 GB, the encoder this graph loads\n- [qwen3.5_9b_w4a8_gptq.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) - 7.34 GB, drop-in swap, saves 2.4 GB\n- [qwen3.5_9b_bf16.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/text_encoders/qwen3.5_9b_bf16.safetensors) - 19.31 GB, full precision\n\n**vae**\n\n- [flux2-vae_small.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae_small.safetensors) - 249 MB, the one this graph loads\n- [flux2-vae.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/vae/flux2-vae.safetensors) - 336 MB, full size, pixel-level differences only\n\n**loras**\n\n- [moss_image_color_depth_control.safetensors](https://huggingface.co/Comfy-Org/MOSS-Image/resolve/main/loras/moss_image_color_depth_control.safetensors) - 313 MB, `LoraLoaderModelOnly` at strength 1.0\n\n**geometry_estimation**\n\n- [moge_3_vitg_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors) - 2.50 GB, the depth estimator used by the subgraph\n- [moge_3_vitl_fp16.safetensors](https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitl_fp16.safetensors) - 741 MB, the lighter ViT-L variant\n\n**Disk footprint**: the graph as shipped needs **18.23 GB** (18,226,010,396 bytes) (5.41 + 9.75 + 0.25 + 0.31 + 2.50). Swapping the text encoder to `w4a8_gptq` brings it to **15.82 GB**; a bf16 checkpoint and encoder instead push it past 32 GB.\n\n## Model Storage Location\n\n```\nšŸ“‚ ComfyUI/\nā”œā”€ā”€ šŸ“‚ models/\nā”‚ ā”œā”€ā”€ šŸ“‚ diffusion_models/\nā”‚ ā”‚ ā””ā”€ā”€ moss_image_sft_int8_convrot.safetensors 5.41 GB\nā”‚ ā”œā”€ā”€ šŸ“‚ text_encoders/\nā”‚ ā”‚ ā””ā”€ā”€ qwen3.5_9b_int8_convrot.safetensors 9.75 GB\nā”‚ ā”œā”€ā”€ šŸ“‚ vae/\nā”‚ ā”‚ ā””ā”€ā”€ flux2-vae_small.safetensors 249 MB\nā”‚ ā”œā”€ā”€ šŸ“‚ loras/\nā”‚ ā”‚ ā””ā”€ā”€ moss_image_color_depth_control.safetensors 313 MB\nā”‚ ā””ā”€ā”€ šŸ“‚ geometry_estimation/\nā”‚ ā””ā”€ā”€ moge_3_vitg_fp16.safetensors 2.50 GB\n```\n\nCore support (`MOSSImageControl`, MoGe nodes) ships from ComfyUI **0.43.0**.\n\n## Report Issue\n\nNote: Please update ComfyUI first ([guide](https://docs.comfy.org/installation/update_comfyui)) and prepare required models. Desktop/Cloud updates follow stable releases, so some nightly-supported models may not be available yet.\n\n- Cannot run / runtime errors: [ComfyUI/issues](https://github.com/comfyanonymous/ComfyUI/issues)\n- UI / frontend issues: [ComfyUI_frontend/issues](https://github.com/Comfy-Org/ComfyUI_frontend/issues)\n- Workflow issues: [workflow_templates/issues](https://github.com/Comfy-Org/workflow_templates/issues)\n" }, "color": "#222", "bgcolor": "#000" }, { "id": 85, "type": "SaveImage", "pos": [ 2450, -180 ], "size": [ 420, 470 ], "flags": {}, "order": 25, "mode": 0, "inputs": [ { "name": "images", "type": "IMAGE", "link": 151 } ], "outputs": [ { "name": "images", "type": "IMAGE", "links": null } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "SaveImage" }, "widgets_values": [ "moss-image/depth-knight-without-lora" ], "widgets_values_named": { "filename_prefix": "image_moss_image_color_depth_control" } }, { "id": 86, "type": "MarkdownNote", "pos": [ -440, 1730 ], "size": [ 4180, 70 ], "flags": {}, "order": 8, "mode": 0, "inputs": [], "outputs": [], "title": "What this test changes", "properties": {}, "widgets_values": [ "Official Comfy-Org MOSS color/depth control template; text encoder swapped to the installed w4a8 one. Color/depth control LoRA bypassed (the template notes the node is a no-op without it)." ], "widgets_values_named": { "text": "Official Comfy-Org MOSS color/depth control template; text encoder swapped to the installed w4a8 one. Color/depth control LoRA bypassed (the template notes the node is a no-op without it)." }, "color": "#432", "bgcolor": "#653" } ], "links": [ [ 3, 3, 0, 5, 0, "CLIP" ], [ 4, 3, 0, 6, 0, "CLIP" ], [ 9, 4, 0, 9, 1, "VAE" ], [ 25, 20, 0, 9, 0, "LATENT" ], [ 27, 22, 0, 20, 4, "SIGMAS" ], [ 35, 2, 0, 20, 0, "MODEL" ], [ 36, 2, 0, 22, 0, "MODEL" ], [ 59, 16, 0, 36, 0, "MODEL" ], [ 69, 36, 0, 2, 0, "MODEL" ], [ 91, 1, 0, 49, 0, "MODEL" ], [ 94, 21, 0, 20, 3, "SAMPLER" ], [ 99, 5, 0, 53, 0, "CONDITIONING" ], [ 100, 6, 0, 53, 1, "CONDITIONING" ], [ 101, 53, 0, 20, 1, "CONDITIONING" ], [ 102, 53, 1, 20, 2, "CONDITIONING" ], [ 105, 4, 0, 53, 2, "VAE" ], [ 106, 54, 0, 20, 5, "LATENT" ], [ 117, 49, 0, 16, 0, "MODEL" ], [ 119, 9, 0, 42, 0, "IMAGE" ], [ 125, 61, 0, 54, 0, "INT" ], [ 126, 61, 1, 54, 1, "INT" ], [ 138, 55, 0, 77, 0, "IMAGE" ], [ 140, 77, 0, 76, 0, "IMAGE" ], [ 141, 77, 0, 61, 0, "IMAGE" ], [ 147, 76, 0, 79, 0, "IMAGE" ], [ 148, 79, 0, 53, 4, "IMAGE" ], [ 149, 42, 0, 80, 1, "IMAGE" ], [ 150, 79, 0, 80, 0, "IMAGE" ], [ 151, 9, 0, 85, 0, "IMAGE" ] ], "groups": [ { "id": 1, "title": "Models", "bounding": [ -460, -570, 550, 950 ], "color": "#3f789e", "flags": {} }, { "id": 2, "title": "Conditioning", "bounding": [ 120, -140, 970, 520 ], "color": "#A88", "flags": {} }, { "id": 3, "title": "Group", "bounding": [ 120, -570, 970, 400 ], "color": "#3f789e", "flags": {} }, { "id": 4, "title": "Sampling", "bounding": [ 1120, -570, 740, 950 ], "color": "#3f789e", "flags": {} }, { "id": 5, "title": "Control Images", "bounding": [ -460, 410, 1550, 640 ], "color": "#A88", "flags": {} }, { "id": 6, "title": "Image Size", "bounding": [ 1120, 410, 740, 640 ], "color": "#3f789e", "flags": {} } ], "definitions": { "subgraphs": [ { "id": "bd7417be-5473-468c-99f1-78759bff8966", "version": 1, "state": { "lastGroupId": 6, "lastNodeId": 86, "lastLinkId": 152, "lastRerouteId": 0 }, "revision": 0, "config": {}, "name": "Geometry Estimation (MoGe 3)", "inputNode": { "id": -10, "bounding": [ -4222.843118848306, 5870.2764474883315, 135.919921875, 148 ] }, "outputNode": { "id": -20, "bounding": [ -3020.2786458333335, 5890.562500000001, 128, 68 ] }, "inputs": [ { "id": "79d73cb8-4bea-46a4-b51f-667917823086", "name": "image", "type": "IMAGE", "linkIds": [ 131 ], "localized_name": "image", "pos": [ -4110.923196973306, 5894.2764474883315 ] }, { "id": "7433a2b5-b40a-42d0-84f7-78d05c4fb561", "name": "resolution_level", "type": "INT", "linkIds": [ 133 ], "pos": [ -4110.923196973306, 5914.2764474883315 ] }, { "id": "7d118489-73ef-46ca-b118-26c719144ae9", "name": "batch_size", "type": "INT", "linkIds": [ 136 ], "pos": [ -4110.923196973306, 5934.2764474883315 ] }, { "id": "17b4dad4-b955-4c27-b929-61e89bb70a61", "name": "output", "type": "COMBO", "linkIds": [ 134 ], "pos": [ -4110.923196973306, 5954.2764474883315 ] }, { "id": "99726d76-065b-4f44-abb7-37e47bafe353", "name": "model_name", "type": "COMBO", "linkIds": [ 135 ], "pos": [ -4110.923196973306, 5974.2764474883315 ] } ], "outputs": [ { "id": "5cbd5fcf-5810-4e2d-a5fc-3b4ba3577d9a", "name": "IMAGE", "type": "IMAGE", "linkIds": [ 132 ], "localized_name": "IMAGE", "pos": [ -2996.2786458333335, 5914.562500000001 ] } ], "widgets": [], "nodes": [ { "id": 73, "type": "LoadMoGeModel", "pos": [ -3990, 5840 ], "size": [ 270, 70 ], "flags": {}, "order": 0, "mode": 0, "inputs": [ { "localized_name": "model_name", "name": "model_name", "type": "COMBO", "widget": { "name": "model_name" }, "link": 135 } ], "outputs": [ { "localized_name": "MOGE_MODEL", "name": "MOGE_MODEL", "type": "MOGE_MODEL", "links": [ 129 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "LoadMoGeModel", "models": [ { "name": "moge_3_vitg_fp16.safetensors", "url": "https://huggingface.co/Comfy-Org/MoGe/resolve/main/geometry_estimation/moge_3_vitg_fp16.safetensors", "directory": "geometry_estimation" } ] }, "widgets_values": [ "moge_3_vitg_fp16.safetensors" ], "widgets_values_named": { "model_name": "moge_3_vitg_fp16.safetensors" } }, { "id": 74, "type": "MoGeInference", "pos": [ -3670, 5840 ], "size": [ 270, 200 ], "flags": {}, "order": 1, "mode": 0, "inputs": [ { "localized_name": "moge_model", "name": "moge_model", "type": "MOGE_MODEL", "link": 129 }, { "localized_name": "image", "name": "image", "type": "IMAGE", "link": 131 }, { "localized_name": "resolution_level", "name": "resolution_level", "type": "INT", "widget": { "name": "resolution_level" }, "link": 133 }, { "localized_name": "batch_size", "name": "batch_size", "type": "INT", "widget": { "name": "batch_size" }, "link": 136 } ], "outputs": [ { "localized_name": "moge_geometry", "name": "moge_geometry", "type": "MOGE_GEOMETRY", "links": [ 130 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "MoGeInference" }, "widgets_values": [ 9, 0, 4, true, true, 3 ], "widgets_values_named": { "resolution_level": 9, "fov_x_degrees": 0, "batch_size": 4, "force_projection": true, "apply_mask": true, "refine_steps": 3 } }, { "id": 75, "type": "MoGeRender", "pos": [ -3350, 5840 ], "size": [ 270, 70 ], "flags": {}, "order": 2, "mode": 0, "inputs": [ { "localized_name": "moge_geometry", "name": "moge_geometry", "type": "MOGE_GEOMETRY", "link": 130 }, { "localized_name": "output", "name": "output", "type": "COMBO", "widget": { "name": "output" }, "link": 134 } ], "outputs": [ { "localized_name": "IMAGE", "name": "IMAGE", "type": "IMAGE", "links": [ 132 ] } ], "properties": { "cnr_id": "comfy-core", "ver": "0.43.1", "Node name for S&R": "MoGeRender" }, "widgets_values": [ "mask" ], "widgets_values_named": { "output": "mask" } } ], "groups": [], "links": [ { "id": 129, "origin_id": 73, "origin_slot": 0, "target_id": 74, "target_slot": 0, "type": "MOGE_MODEL" }, { "id": 130, "origin_id": 74, "origin_slot": 0, "target_id": 75, "target_slot": 0, "type": "MOGE_GEOMETRY" }, { "id": 131, "origin_id": -10, "origin_slot": 0, "target_id": 74, "target_slot": 1, "type": "IMAGE" }, { "id": 132, "origin_id": 75, "origin_slot": 0, "target_id": -20, "target_slot": 0, "type": "IMAGE" }, { "id": 133, "origin_id": -10, "origin_slot": 1, "target_id": 74, "target_slot": 2, "type": "INT" }, { "id": 134, "origin_id": -10, "origin_slot": 3, "target_id": 75, "target_slot": 1, "type": "COMBO" }, { "id": 135, "origin_id": -10, "origin_slot": 4, "target_id": 73, "target_slot": 0, "type": "COMBO" }, { "id": 136, "origin_id": -10, "origin_slot": 2, "target_id": 74, "target_slot": 3, "type": "INT" } ], "extra": {} } ] }, "config": {}, "extra": { "ds": { "scale": 0.4226044165468412, "offset": [ 621.6279395380957, 606.7906984884523 ] }, "frontendVersion": "1.57.4", "VHS_latentpreview": false, "VHS_latentpreviewrate": 0, "VHS_MetadataImage": true, "VHS_KeepIntermediate": true }, "version": 0.4 }