valknar/runpod
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: rebuild SD3.5 workflow with TripleCLIPLoader

Browse Source 
SD3.5 checkpoint doesn't contain CLIP encoders. Now using:
- CheckpointLoaderSimple for MODEL and VAE
- TripleCLIPLoader for CLIP-L, CLIP-G, and T5-XXL
- Standard CLIPTextEncode for prompts

This fixes the "clip input is invalid: None" error.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Sebastian Krüger
2025-11-23 10:56:09 +01:00
parent 91ed1aa9e3
commit 5c1e9d092b
1 changed files with 59 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

100
comfyui/workflows/text-to-image/sd35-large-t2i-production-v1.json
View File

@@ -1,12 +1,12 @@
{ {
"last_node_id": 12, "last_node_id": 9,
"last_link_id": 15, "last_link_id": 10,
"nodes": [ "nodes": [
{ {
"id": 1, "id": 1,
"type": "CheckpointLoaderSimple", "type": "CheckpointLoaderSimple",
"pos": [50, 100], "pos": [50, 100],
"size": {"0": 350, "1": 100}, "size": [350, 100],
"flags": {}, "flags": {},
"order": 0, "order": 0,
"mode": 0, "mode": 0,
@@ -20,8 +20,7 @@
{ {
"name": "CLIP", "name": "CLIP",
"type": "CLIP", "type": "CLIP",
"links": [2, 3], "links": null
"slot_index": 1
}, },
{ {
"name": "VAE", "name": "VAE",
@@ -36,19 +35,41 @@
"widgets_values": ["sd3.5_large.safetensors"], "widgets_values": ["sd3.5_large.safetensors"],
"title": "SD3.5 Large Checkpoint Loader" "title": "SD3.5 Large Checkpoint Loader"
}, },
{
"id": 9,
"type": "TripleCLIPLoader",
"pos": [50, 250],
"size": [350, 150],
"flags": {},
"order": 1,
"mode": 0,
"outputs": [
{
"name": "CLIP",
"type": "CLIP",
"links": [9, 10],
"slot_index": 0
}
],
"properties": {
"Node name for S&R": "TripleCLIPLoader"
},
"widgets_values": ["clip_l.safetensors", "clip_g.safetensors", "t5xxl_fp16.safetensors"],
"title": "SD3.5 CLIP Loaders (L+G+T5)"
},
{ {
"id": 2, "id": 2,
"type": "CLIPTextEncode", "type": "CLIPTextEncode",
"pos": [450, 100], "pos": [450, 100],
"size": {"0": 400, "1": 200}, "size": [400, 200],
"flags": {}, "flags": {},
"order": 1, "order": 2,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
"name": "clip", "name": "clip",
"type": "CLIP", "type": "CLIP",
"link": 2 "link": 9
} }
], ],
"outputs": [ "outputs": [
@@ -69,15 +90,15 @@
"id": 3, "id": 3,
"type": "CLIPTextEncode", "type": "CLIPTextEncode",
"pos": [450, 350], "pos": [450, 350],
"size": {"0": 400, "1": 200}, "size": [400, 200],
"flags": {}, "flags": {},
"order": 2, "order": 3,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
"name": "clip", "name": "clip",
"type": "CLIP", "type": "CLIP",
"link": 3 "link": 10
} }
], ],
"outputs": [ "outputs": [
@@ -98,9 +119,9 @@
"id": 4, "id": 4,
"type": "EmptyLatentImage", "type": "EmptyLatentImage",
"pos": [450, 600], "pos": [450, 600],
"size": {"0": 315, "1": 106}, "size": [315, 106],
"flags": {}, "flags": {},
"order": 3, "order": 4,
"mode": 0, "mode": 0,
"outputs": [ "outputs": [
{ {
@@ -120,9 +141,9 @@
"id": 5, "id": 5,
"type": "KSampler", "type": "KSampler",
"pos": [900, 100], "pos": [900, 100],
"size": {"0": 315, "1": 474}, "size": [315, 474],
"flags": {}, "flags": {},
"order": 4, "order": 5,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
@@ -157,24 +178,16 @@
"properties": { "properties": {
"Node name for S&R": "KSampler" "Node name for S&R": "KSampler"
}, },
"widgets_values": [ "widgets_values": [42, "fixed", 28, 4.5, "dpmpp_2m", "karras", 1],
42,
"fixed",
28,
4.5,
"dpmpp_2m",
"karras",
1
],
"title": "SD3.5 Sampler (28 steps)" "title": "SD3.5 Sampler (28 steps)"
}, },
{ {
"id": 6, "id": 6,
"type": "VAEDecode", "type": "VAEDecode",
"pos": [1270, 100], "pos": [1270, 100],
"size": {"0": 210, "1": 46}, "size": [210, 46],
"flags": {}, "flags": {},
"order": 5, "order": 6,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
@@ -192,7 +205,7 @@
{ {
"name": "IMAGE", "name": "IMAGE",
"type": "IMAGE", "type": "IMAGE",
"links": [9, 10], "links": [2, 3],
"slot_index": 0 "slot_index": 0
} }
], ],
@@ -205,15 +218,15 @@
"id": 7, "id": 7,
"type": "PreviewImage", "type": "PreviewImage",
"pos": [1530, 100], "pos": [1530, 100],
"size": {"0": 400, "1": 400}, "size": [400, 400],
"flags": {}, "flags": {},
"order": 6, "order": 7,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
"name": "images", "name": "images",
"type": "IMAGE", "type": "IMAGE",
"link": 9 "link": 2
} }
], ],
"properties": { "properties": {
@@ -225,15 +238,15 @@
"id": 8, "id": 8,
"type": "SaveImage", "type": "SaveImage",
"pos": [1530, 550], "pos": [1530, 550],
"size": {"0": 400, "1": 100}, "size": [400, 100],
"flags": {}, "flags": {},
"order": 7, "order": 8,
"mode": 0, "mode": 0,
"inputs": [ "inputs": [
{ {
"name": "images", "name": "images",
"type": "IMAGE", "type": "IMAGE",
"link": 10 "link": 3
} }
], ],
"properties": { "properties": {
@@ -245,28 +258,27 @@
], ],
"links": [ "links": [
[1, 1, 0, 5, 0, "MODEL"], [1, 1, 0, 5, 0, "MODEL"],
[2, 1, 1, 2, 0, "CLIP"], [2, 6, 0, 7, 0, "IMAGE"],
[3, 1, 1, 3, 0, "CLIP"], [3, 6, 0, 8, 0, "IMAGE"],
[4, 1, 2, 6, 1, "VAE"], [4, 1, 2, 6, 1, "VAE"],
[5, 2, 0, 5, 1, "CONDITIONING"], [5, 2, 0, 5, 1, "CONDITIONING"],
[6, 3, 0, 5, 2, "CONDITIONING"], [6, 3, 0, 5, 2, "CONDITIONING"],
[7, 4, 0, 5, 3, "LATENT"], [7, 4, 0, 5, 3, "LATENT"],
[8, 5, 0, 6, 0, "LATENT"], [8, 5, 0, 6, 0, "LATENT"],
[9, 6, 0, 7, 0, "IMAGE"], [9, 9, 0, 2, 0, "CLIP"],
[10, 6, 0, 8, 0, "IMAGE"] [10, 9, 0, 3, 0, "CLIP"]
], ],
"groups": [], "groups": [],
"config": {}, "config": {},
"extra": { "extra": {
"workflow_info": { "workflow_info": {
"name": "Stable Diffusion 3.5 Large Text-to-Image Production", "name": "Stable Diffusion 3.5 Large Text-to-Image Production",
"version": "1.0.0", "version": "1.2.0",
"author": "RunPod AI Model Orchestrator",
"description": "Latest generation text-to-image using Stable Diffusion 3.5 Large (28 steps). Provides excellent photorealism and prompt adherence.", "description": "Latest generation text-to-image using Stable Diffusion 3.5 Large (28 steps). Provides excellent photorealism and prompt adherence.",
"category": "text-to-image", "category": "text-to-image",
"tags": ["sd3.5", "stable-diffusion", "large", "production", "t2i", "photorealistic"], "tags": ["sd3.5", "stable-diffusion", "large", "production", "t2i", "photorealistic"],
"requirements": { "requirements": {
"models": ["stable-diffusion-3.5-large"], "models": ["SD3.5-large", "CLIP-L", "CLIP-G", "T5-XXL FP16"],
"custom_nodes": [], "custom_nodes": [],
"vram_min": "18GB", "vram_min": "18GB",
"vram_recommended": "24GB" "vram_recommended": "24GB"
@@ -351,7 +363,13 @@
"avg_generation_time": "35-45 seconds", "avg_generation_time": "35-45 seconds",
"vram_usage": "~18-20GB", "vram_usage": "~18-20GB",
"gpu_utilization": "95-100%" "gpu_utilization": "95-100%"
} },
"notes": [
"SD3.5 uses CheckpointLoaderSimple for MODEL+VAE",
"TripleCLIPLoader loads CLIP-L, CLIP-G, and T5-XXL separately",
"Checkpoint file doesn't include CLIP encoders",
"All three CLIP encoders required for best quality"
]
} }
}, },
"version": 0.4 "version": 0.4