Organize optimized workflows by reusable type

This commit is contained in:
Morpheus
2026-09-23 08:47:12 +00:00
parent a321551449
commit 7ff0b7547a
30 changed files with 88 additions and 22 deletions
+3
View File
@@ -0,0 +1,3 @@
# I2I workflows
Use [`flux/`](flux/) or [`qwen/`](qwen/) for one-reference-image transformations. The workflows infer the subject from the input image instead of naming a specific character or scene.
@@ -1,4 +1,4 @@
# Flux.2 Klein — photorealistic character conversion
# Flux.2 Klein — optimized I2I
Converts any supplied reference image into a photorealistic image of the same subject. The prompt is intentionally generic: identity, pose, clothing, accessories, colors, proportions, composition, and framing are inferred from the reference image rather than hard-coded.
+23
View File
@@ -0,0 +1,23 @@
# Qwen Image 2.1 — optimized I2I
Converts any supplied reference image into a photorealistic image of the same subject. The prompt is generic: identity, pose, clothing, accessories, colors, proportions, composition, and framing are inferred from the input rather than hard-coded.
## Reusable inputs
- `input.png` in node `4` — replace with the reference image.
- Node `7` positive and negative prompts — adapt the transformation without describing a specific subject.
- Node `10` seed — change for alternate results or keep fixed for comparisons.
- Nodes `5` and `14` — change the generation and output dimensions. The default is portrait 1080×1920 for person-focused images.
## Optimization baseline
- Model: `qwen_image_2.1_int8_convrot.safetensors`
- Sampler: Euler CFG++
- Scheduler: beta
- Steps: 40
- CFG: 3.5
- Measured runtime: approximately 81 seconds on the RTX 3070
The workflow uses the installed Qwen Image 2.1 stack and the flattened `images.image_1` reference input required by the API workflow.
- `workflow.json` — API-format workflow.
+17
View File
@@ -0,0 +1,17 @@
{
"1": {"class_type":"UNETLoader","inputs":{"unet_name":"qwen_image_2.1_int8_convrot.safetensors","weight_dtype":"default"}},
"2": {"class_type":"CLIPLoader","inputs":{"clip_name":"qwen3vl_8b_w4a8.safetensors","type":"qwen_image","device":"default"}},
"3": {"class_type":"VAELoader","inputs":{"vae_name":"qwen_image_2.1_vae_bf16.safetensors"}},
"4": {"class_type":"LoadImage","inputs":{"image":"input.png"}},
"5": {"class_type":"ResizeAndPadImage","inputs":{"image":["4",0],"target_width":576,"target_height":1024,"padding_color":"white","interpolation":"lanczos"}},
"6": {"class_type":"QwenImage21Cache","inputs":{"model":["1",0],"device":"auto","dtype":"int8"}},
"7": {"class_type":"TextEncodeQwenImage21","inputs":{"clip":["2",0],"prompt":"Transform the supplied reference image into a photorealistic image of the same subject. Infer identity, facial features, skin tone, hairstyle, clothing, accessories, colors, pose, body proportions, composition, and framing directly from the reference. Preserve those visual attributes faithfully while replacing the original rendering style with natural human anatomy, realistic skin and hair texture, physically accurate materials, believable fabric and surface detail, natural hands and fingers, cinematic soft light, realistic shadows, coherent perspective, premium editorial photography, high dynamic range, photorealistic, highly detailed. Do not redesign, restyle, substitute, add, remove, mirror, or crop the subject. No text or logos.","negative_prompt":"different person, changed identity, changed face, changed hairstyle, changed clothing, changed colors, altered accessories, changed pose, different composition, extra person, duplicate person, extra limbs, malformed hands, fused fingers, distorted anatomy, cropped head, cropped feet, mirrored image, cartoon, illustration, anime, CGI, plastic skin, waxy skin, blurry, low detail, text, logo, watermark","resolution":0,"images.image_1":["5",0],"vae":["3",0]}},
"8": {"class_type":"KSamplerSelect","inputs":{"sampler_name":"euler_cfg_pp"}},
"9": {"class_type":"BasicScheduler","inputs":{"model":["6",0],"scheduler":"beta","steps":40,"denoise":1}},
"10": {"class_type":"RandomNoise","inputs":{"noise_seed":20260921}},
"11": {"class_type":"CFGGuider","inputs":{"model":["6",0],"positive":["7",0],"negative":["7",1],"cfg":3.5}},
"12": {"class_type":"SamplerCustomAdvanced","inputs":{"noise":["10",0],"guider":["11",0],"sampler":["8",0],"sigmas":["9",0],"latent_image":["7",2]}},
"13": {"class_type":"VAEDecode","inputs":{"samples":["12",0],"vae":["3",0]}},
"14": {"class_type":"ResizeAndPadImage","inputs":{"image":["13",0],"target_width":1080,"target_height":1920,"padding_color":"white","interpolation":"lanczos"}},
"15": {"class_type":"SaveImage","inputs":{"images":["14",0],"filename_prefix":"fluxExperiments/imageEdit/qwen_i2i"}}
}
+3
View File
@@ -0,0 +1,3 @@
# MultipleRef2I workflows
Use [`flux/`](flux/) for the current best multi-reference image composition workflow. It accepts a primary subject reference and an environment or scene reference. A Qwen MultipleRef2I workflow has not yet been created.
@@ -1,4 +1,4 @@
# Flux.2 Klein 9B multi-reference composite
# Flux.2 Klein — optimized MultipleRef2I
API workflow for combining two reference images with Flux.2 Klein:
+32 -8
View File
@@ -1,13 +1,37 @@
# Optimized workflows
This directory contains the current tested production-oriented workflows for the available hardware.
This directory contains the best tested workflows for the available hardware. Choose the workflow type first, then choose the model implementation.
## Current baseline
## Workflow types
- **Flux.2 Klein 9B FP8** text-to-image at 1920×1080
- Euler CFG++ sampler, 6 steps, CFG 1.3
- Fixed seed `20260920` for repeatable baseline comparisons
- Measured generation time: approximately 78 seconds on the RTX 3070
- Optional SeedVR2 3B FP8 second pass tested separately at 2560×1440; the full 3840×2160 target exceeded 8 GB VRAM
- [`T2I/`](T2I/) — text-to-image generation.
- [`flux/`](T2I/flux/) — Flux.2 Klein best baseline.
- [`qwen/`](T2I/qwen/) — Qwen Image 2.1 best baseline.
- [`I2I/`](I2I/) — one-reference-image transformation or editing.
- [`flux/`](I2I/flux/) — Flux.2 Klein photorealistic conversion.
- [`qwen/`](I2I/qwen/) — Qwen Image 2.1 photorealistic conversion.
- [`MultipleRef2I/`](MultipleRef2I/) — multi-reference image composition.
- [`flux/`](MultipleRef2I/flux/) — Flux.2 Klein best multi-reference workflow.
- Qwen MultipleRef2I is not yet available.
See [`text-to-image/flux2klein-9b-3070-best/`](text-to-image/flux2klein-9b-3070-best/).
## Reusable variables
These are the values to change for a specific task:
- Prompt and negative prompt
- Seed
- Input image or images
- Generation and output resolution
## Optimization variables
These are the values used to balance speed and quality:
- Step count
- Sampler
- Scheduler
- CFG value
The workflow files retain the tested baseline values. Each workflow README documents its model, inputs, dimensions, and measured runtime.
The previous prompt-specific and specialized workflows are retained under [`archive/`](archive/) for reference.
+3
View File
@@ -0,0 +1,3 @@
# T2I workflows
Use [`flux/`](flux/) or [`qwen/`](qwen/) for the best tested text-to-image baseline. Prompts, seeds, and dimensions are intended to be replaced per task; sampler, scheduler, CFG, and steps are the optimization controls.
@@ -1,4 +1,4 @@
# Flux.2 Klein 9B — RTX 3070 best baseline
# Flux.2 Klein 9B — optimized T2I
This is the current best tested Flux.2 Klein text-to-image workflow for the 8 GB RTX 3070.
@@ -19,7 +19,7 @@ Files:
- Seed: `20260920`
- Measured generation time: approximately 78 seconds
The seed and prompt are intentionally fixed so future sampler, scheduler, step, CFG, LoRA, and second-pass experiments can be compared consistently.
The included seed and prompt are a controlled baseline example. For reuse, replace the prompt, seed, and resolution; sampler, scheduler, steps, and CFG are the optimization controls.
## Optional second pass
@@ -1,6 +1,6 @@
# Qwen Image 2.1 — spaceship canteen baseline
# Qwen Image 2.1 — optimized T2I
Qwen Image 2.1 text-to-image workflow using the same spaceship-canteen prompt as the Flux.2 Klein baseline.
Qwen Image 2.1 text-to-image workflow using the best tested sampler, scheduler, step, and CFG balance. The prompt is an editable example rather than a workflow identity.
## Settings
@@ -16,9 +16,8 @@ Qwen Image 2.1 text-to-image workflow using the same spaceship-canteen prompt as
- Seed: 20260920
- Warm generation time: approximately 22 seconds on the server RTX 3070
The positive and negative prompts are intentionally identical to the Flux baseline for controlled comparison. The final crop removes the white side bands produced by padding the square Qwen output.
For reuse, replace the positive and negative prompts, seed, and output dimensions. The final crop removes the white side bands produced by padding the square Qwen output.
Files:
- workflow.json — API format
- ui-workflow.json — ComfyUI web UI format
-3
View File
@@ -1,3 +0,0 @@
# Image-edit workflows
The current image-edit baseline is the Flux.2 Klein photorealistic character conversion workflow.
-3
View File
@@ -1,3 +0,0 @@
# Text-to-image
The optimized text-to-image collection currently focuses on the tested Flux.2 Klein baseline.