Files
comfyui-workflows/optimized/T2I/qwen/README.md
T

129 lines
8.2 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Qwen Image 2.1 — optimized T2I
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
- Model: qwen_image_2.1_int8_convrot.safetensors
- Text encoder: qwen3vl_8b_w4a8.safetensors
- VAE: qwen_image_2.1_vae_bf16.safetensors
- Internal generation: 1088×1920 portrait latent (generated directly at target resolution)
- Final output: 1088×1920, 9:16, decoded directly from the target-resolution latent
- Sampler: res_multistep
- Scheduler: beta
- Steps: 25
- CFG: 2.5
- Seed: 20260920
- Runtime: 110.2s for the current best; recorded per test in the matrix
For reuse, replace the positive and negative prompts, seed, and target resolution. The latent and decoded image now use the same 1088×1920 dimensions.
## Planned optimization matrix
Keep the prompt, seed, and resolution from [`../variables.json`](../variables.json) fixed. Update each row only after its exact configuration has been run and evaluated.
| ID | CFG | Steps | Sampler | Scheduler | Runtime (s) | Tested? | Current Best |
|---|---:|---:|---|---|---:|---|---|
| Q-T2I-01 | 2.5 | 12 | euler_cfg_pp | simple | 315.5 | Yes | No |
| Q-T2I-02 | 1.0 | 12 | euler_cfg_pp | simple | 290.6 | Yes | No |
| Q-T2I-03 | 1.1 | 12 | euler_cfg_pp | simple | 94.7 | Yes | No |
| Q-T2I-04 | 1.2 | 12 | euler_cfg_pp | simple | 53.0 | Yes | No |
| Q-T2I-05 | 1.3 | 12 | euler_cfg_pp | simple | 53.1 | Yes | No |
| Q-T2I-06 | 1.4 | 12 | euler_cfg_pp | simple | 56.1 | Yes | No |
| Q-T2I-07 | 1.5 | 12 | euler_cfg_pp | simple | 56.4 | Yes | No |
| Q-T2I-08 | 1.6 | 12 | euler_cfg_pp | simple | 55.2 | Yes | No |
| Q-T2I-09 | 1.8 | 12 | euler_cfg_pp | simple | 53.4 | Yes | No |
| Q-T2I-10 | 2.0 | 12 | euler_cfg_pp | simple | 53.6 | Yes | No |
| Q-T2I-11 | 2.2 | 12 | euler_cfg_pp | simple | 55.9 | Yes | No |
| Q-T2I-12 | 2.4 | 12 | euler_cfg_pp | simple | 54.7 | Yes | No |
| Q-T2I-13 | 2.5 | 12 | euler | simple | 54.3 | Yes | No |
| Q-T2I-14 | 2.5 | 12 | euler_ancestral | simple | 54.4 | Yes | No |
| Q-T2I-15 | 2.5 | 12 | euler_ancestral_cfg_pp | simple | 54.2 | Yes | No |
| Q-T2I-16 | 2.5 | 12 | res_multistep | simple | 57.5 | Yes | No |
| Q-T2I-17 | 2.5 | 12 | res_multistep_cfg_pp | simple | 54.3 | Yes | No |
| Q-T2I-18 | 2.5 | 12 | res_multistep_ancestral | simple | 56.8 | Yes | No |
| Q-T2I-19 | 2.5 | 12 | res_multistep_ancestral_cfg_pp | simple | 57.2 | Yes | No |
| Q-T2I-20 | 2.5 | 12 | euler_cfg_pp | beta | 54.1 | Yes | No |
| Q-T2I-21 | 2.5 | 12 | euler_cfg_pp | sgm_uniform | 54.5 | Yes | No |
| Q-T2I-22 | 2.5 | 12 | euler_cfg_pp | karras | 55.3 | Yes | No |
| Q-T2I-23 | 2.5 | 12 | euler_cfg_pp | exponential | 54.1 | Yes | No |
| Q-T2I-24 | 2.5 | 12 | euler_cfg_pp | normal | 54.4 | Yes | No |
| Q-T2I-25 | 2.5 | 20 | euler_cfg_pp | beta | 89.7 | Yes | No |
| Q-T2I-26 | 2.5 | 30 | euler_cfg_pp | beta | 131.6 | Yes | No |
| Q-T2I-27 | 2.5 | 40 | euler_cfg_pp | beta | 175.4 | Yes | No |
| Q-T2I-28 | 2.5 | 20 | euler_ancestral | beta | 89.8 | Yes | No |
| Q-T2I-29 | 2.5 | 20 | res_multistep | beta | 88.4 | Yes | No |
| Q-T2I-30 | 2.5 | 25 | euler_cfg_pp | simple | 159.3 | Yes | No |
| Q-T2I-31 | 1.0 | 25 | euler_cfg_pp | simple | 108.5 | Yes | No |
| Q-T2I-32 | 1.1 | 25 | euler_cfg_pp | simple | 109.4 | Yes | No |
| Q-T2I-33 | 1.2 | 25 | euler_cfg_pp | simple | 109.3 | Yes | No |
| Q-T2I-34 | 1.3 | 25 | euler_cfg_pp | simple | 110.0 | Yes | No |
| Q-T2I-35 | 1.4 | 25 | euler_cfg_pp | simple | 109.7 | Yes | No |
| Q-T2I-36 | 1.5 | 25 | euler_cfg_pp | simple | 109.9 | Yes | No |
| Q-T2I-37 | 1.6 | 25 | euler_cfg_pp | simple | 110.0 | Yes | No |
| Q-T2I-38 | 1.8 | 25 | euler_cfg_pp | simple | — | Yes | No |
| Q-T2I-39 | 2.0 | 25 | euler_cfg_pp | simple | 111.4 | Yes | No |
| Q-T2I-40 | 2.2 | 25 | euler_cfg_pp | simple | 109.9 | Yes | No |
| Q-T2I-41 | 2.4 | 25 | euler_cfg_pp | simple | 110.5 | Yes | No |
| Q-T2I-42 | 2.5 | 25 | euler | simple | 110.7 | Yes | No |
| Q-T2I-43 | 2.5 | 25 | euler_ancestral | simple | 113.3 | Yes | No |
| Q-T2I-44 | 2.5 | 25 | euler_ancestral_cfg_pp | simple | 110.1 | Yes | No |
| Q-T2I-45 | 2.5 | 25 | res_multistep | simple | 109.8 | Yes | No |
| Q-T2I-46 | 2.5 | 25 | res_multistep_cfg_pp | simple | 110.0 | Yes | No |
| Q-T2I-47 | 2.5 | 25 | res_multistep_ancestral | simple | 111.4 | Yes | No |
| Q-T2I-48 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | simple | 111.1 | Yes | No |
| Q-T2I-49 | 2.5 | 25 | euler_cfg_pp | beta | 110.6 | Yes | No |
| Q-T2I-50 | 2.5 | 25 | euler_cfg_pp | sgm_uniform | 109.8 | Yes | No |
| Q-T2I-51 | 2.5 | 25 | euler_cfg_pp | karras | 110.9 | Yes | No |
| Q-T2I-52 | 2.5 | 25 | euler_cfg_pp | exponential | 113.2 | Yes | No |
| Q-T2I-53 | 2.5 | 25 | euler_cfg_pp | normal | 114.3 | Yes | No |
| Q-T2I-54 | 2.5 | 25 | euler | beta | 145.6 | Yes | No |
| Q-T2I-55 | 2.5 | 25 | euler_ancestral | beta | 107.7 | Yes | No |
| Q-T2I-56 | 2.5 | 25 | euler_ancestral_cfg_pp | beta | 109.6 | Yes | No |
| Q-T2I-57 | 2.5 | 25 | res_multistep | beta | 110.2 | Yes | Yes |
| Q-T2I-58 | 2.5 | 25 | res_multistep_cfg_pp | beta | 109.4 | Yes | No |
| Q-T2I-59 | 2.5 | 25 | res_multistep_ancestral | beta | 109.4 | Yes | No |
| Q-T2I-60 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | beta | 110.8 | Yes | No |
| Q-T2I-61 | 2.5 | 25 | euler | sgm_uniform | 114.2 | Yes | No |
| Q-T2I-62 | 2.5 | 25 | euler_ancestral | sgm_uniform | 110.2 | Yes | No |
| Q-T2I-63 | 2.5 | 25 | euler_ancestral_cfg_pp | sgm_uniform | 110.1 | Yes | No |
| Q-T2I-64 | 2.5 | 25 | res_multistep | sgm_uniform | 111.3 | Yes | No |
| Q-T2I-65 | 2.5 | 25 | res_multistep_cfg_pp | sgm_uniform | 109.9 | Yes | No |
| Q-T2I-66 | 2.5 | 25 | res_multistep_ancestral | sgm_uniform | 111.3 | Yes | No |
| Q-T2I-67 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | sgm_uniform | 108.8 | Yes | No |
| Q-T2I-68 | 2.5 | 25 | euler | karras | 110.0 | Yes | No |
| Q-T2I-69 | 2.5 | 25 | euler_ancestral | karras | 109.4 | Yes | No |
| Q-T2I-70 | 2.5 | 25 | euler_ancestral_cfg_pp | karras | 108.8 | Yes | No |
| Q-T2I-71 | 2.5 | 25 | res_multistep | karras | 108.6 | Yes | No |
| Q-T2I-72 | 2.5 | 25 | res_multistep_cfg_pp | karras | 115.2 | Yes | No |
| Q-T2I-73 | 2.5 | 25 | res_multistep_ancestral | karras | 109.5 | Yes | No |
| Q-T2I-74 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | karras | 112.5 | Yes | No |
| Q-T2I-75 | 2.5 | 25 | euler | exponential | 109.8 | Yes | No |
| Q-T2I-76 | 2.5 | 25 | euler_ancestral | exponential | 109.0 | Yes | No |
| Q-T2I-77 | 2.5 | 25 | euler_ancestral_cfg_pp | exponential | 111.5 | Yes | No |
| Q-T2I-78 | 2.5 | 25 | res_multistep | exponential | 108.7 | Yes | No |
| Q-T2I-79 | 2.5 | 25 | res_multistep_cfg_pp | exponential | 109.1 | Yes | No |
| Q-T2I-80 | 2.5 | 25 | res_multistep_ancestral | exponential | 109.1 | Yes | No |
| Q-T2I-81 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | exponential | 109.9 | Yes | No |
| Q-T2I-82 | 2.5 | 25 | euler | normal | 108.9 | Yes | No |
| Q-T2I-83 | 2.5 | 25 | euler_ancestral | normal | 112.5 | Yes | No |
| Q-T2I-84 | 2.5 | 25 | euler_ancestral_cfg_pp | normal | 111.4 | Yes | No |
| Q-T2I-85 | 2.5 | 25 | res_multistep | normal | 109.5 | Yes | No |
| Q-T2I-86 | 2.5 | 25 | res_multistep_cfg_pp | normal | 111.9 | Yes | No |
| Q-T2I-87 | 2.5 | 25 | res_multistep_ancestral | normal | 109.5 | Yes | No |
| Q-T2I-88 | 2.5 | 25 | res_multistep_ancestral_cfg_pp | normal | 111.9 | Yes | No |
Files:
- workflow.json — API format
- ui-workflow.json — ComfyUI web UI format
Visual evaluation notes:
- Full matrix status: 88/88 tested; 35/35 newly collected in this batch.
- Current best: Q-T2I-57: `res_multistep`, beta, 25 steps, CFG 2.5, 110.2s. User review rated it highest so far for skin texture and foot anatomy.
- Scheduler comparison note: at CFG 1.0 with Res Multistep and 25 steps, beta produced more realistic skin than simple; beta remains preferred for the production baseline.
- Strong new alternatives, but not promoted: Q-T2I-54–57 and Q-T2I-61–64. Q-T2I-82 and Q-T2I-84 initially appeared strong technically, but were rejected after review for visibly deformed legs/feet.
- Valid but weaker: Q-T2I-85.
- Rejected for anatomy, severe corruption, or unusable exposure: Q-T2I-82, Q-T2I-84, Q-T2I-58–60, Q-T2I-65–81, and Q-T2I-86–88.
- Earlier rejected rows remain: Q-T2I-17, Q-T2I-18, Q-T2I-19, Q-T2I-22, Q-T2I-23, Q-T2I-46, Q-T2I-47, Q-T2I-48, Q-T2I-51, and Q-T2I-52.