Reduce Flux 3 Klein T2I test plan
This commit is contained in:
@@ -1,7 +1,7 @@
|
||||
# FLUX 3 Image / Klein
|
||||
|
||||
This is the canonical local FLUX.3 Klein text-to-image workflow and clean
|
||||
sampler/scheduler test plan. Reference-based tests belong in
|
||||
This is the canonical local FLUX.3 Klein text-to-image workflow and focused
|
||||
optimization plan. Reference-based tests belong in
|
||||
`optimized/I2I/flux3klein/` and `optimized/MultipleRef2I/flux3klein/`.
|
||||
|
||||
No results have been retained for this fresh run. The workflow does not use
|
||||
@@ -17,28 +17,42 @@ the remote `Flux3ImageNode` API.
|
||||
- Resolution: **1088×1920**
|
||||
- References: none
|
||||
- Seed: `310001`
|
||||
- Steps: `4`
|
||||
- Flux Guidance: `3.5`
|
||||
- Scheduler: `BasicScheduler` with `simple`
|
||||
|
||||
The graph uses `FluxGuidance`, not `CFGGuider`; therefore this T2I suite
|
||||
records Flux Guidance rather than calling it CFG. Prompt, resolution, seed,
|
||||
steps, guidance, and reference count stay fixed while sampler and scheduler
|
||||
vary.
|
||||
The graph uses `FluxGuidance`, not `CFGGuider`; this suite records Flux
|
||||
Guidance rather than calling it CFG. Prompt, resolution, seed, guidance,
|
||||
scheduler, and reference count stay fixed unless a row explicitly changes
|
||||
steps or sampler.
|
||||
|
||||
## Fresh test matrix
|
||||
## Focused optimization matrix
|
||||
|
||||
The matrix contains **48 samplers × 9 schedulers = 432 planned runs**. All
|
||||
rows are currently **Planned**. The complete sampler and scheduler lists are
|
||||
stored in `variables.json`; each executed cell must receive a result record in
|
||||
`results/` with its exact settings, runtime, output path, dimensions, and
|
||||
error or visual notes.
|
||||
This follows the practical 18-row scope used by the Flux.2 Klein plan: sampler
|
||||
comparisons first, followed by step-count comparisons. All rows are currently
|
||||
**Planned**.
|
||||
|
||||
| Matrix | Resolution | References | Seed | Guidance | Steps | Samplers | Schedulers | Status |
|
||||
|---|---:|---:|---:|---:|---:|---:|---:|---|
|
||||
| F3I-MATRIX | 1088×1920 | 0 | 310001 | 3.5 | 4 | 48 | 9 | Planned (0/432) |
|
||||
| ID | Guidance | Steps | Sampler | Scheduler | Status | Purpose |
|
||||
|---|---:|---:|---|---|---|---|
|
||||
| F3I-T2I-01 | 3.5 | 4 | euler | simple | Planned | Baseline |
|
||||
| F3I-T2I-02 | 3.5 | 4 | euler_cfg_pp | simple | Planned | CFG++ sampler |
|
||||
| F3I-T2I-03 | 3.5 | 4 | euler_ancestral | simple | Planned | Ancestral sampler |
|
||||
| F3I-T2I-04 | 3.5 | 4 | euler_ancestral_cfg_pp | simple | Planned | Ancestral CFG++ sampler |
|
||||
| F3I-T2I-05 | 3.5 | 4 | res_multistep | simple | Planned | RES multistep sampler |
|
||||
| F3I-T2I-06 | 3.5 | 4 | res_multistep_cfg_pp | simple | Planned | RES multistep CFG++ sampler |
|
||||
| F3I-T2I-07 | 3.5 | 4 | dpmpp_2m | simple | Planned | DPM++ sampler |
|
||||
| F3I-T2I-08 | 3.5 | 4 | ddim | simple | Planned | DDIM sampler |
|
||||
| F3I-T2I-09 | 3.5 | 4 | uni_pc | simple | Planned | UniPC sampler |
|
||||
| F3I-T2I-10 | 3.5 | 2 | euler_cfg_pp | simple | Planned | Low-step comparison |
|
||||
| F3I-T2I-11 | 3.5 | 3 | euler_cfg_pp | simple | Planned | Low-step comparison |
|
||||
| F3I-T2I-12 | 3.5 | 5 | euler_cfg_pp | simple | Planned | Step-count comparison |
|
||||
| F3I-T2I-13 | 3.5 | 6 | euler_cfg_pp | simple | Planned | Step-count comparison |
|
||||
| F3I-T2I-14 | 3.5 | 7 | euler_cfg_pp | simple | Planned | Step-count comparison |
|
||||
| F3I-T2I-15 | 3.5 | 8 | euler_cfg_pp | simple | Planned | Step-count comparison |
|
||||
| F3I-T2I-16 | 3.5 | 4 | heun | simple | Planned | Heun sampler |
|
||||
| F3I-T2I-17 | 3.5 | 4 | dpmpp_sde | simple | Planned | DPM++ SDE sampler |
|
||||
| F3I-T2I-18 | 3.5 | 4 | dpmpp_2m_cfg_pp | simple | Planned | DPM++ CFG++ sampler |
|
||||
|
||||
Do not silently substitute a rejected sampler or scheduler. Record failures
|
||||
as results, preserve the fixed comparison inputs, and update this table only
|
||||
from the actual result records. The first successful `euler + simple` cell is
|
||||
the compatibility baseline; no visual winner should be declared until the
|
||||
outputs have been reviewed under the same prompt and resolution.
|
||||
Record runtime, output dimensions, output path, exact settings, and visual
|
||||
notes for every row under `results/`. Record failures rather than silently
|
||||
substituting another sampler. The first successful baseline is a compatibility
|
||||
reference, not automatically the visual winner.
|
||||
|
||||
Reference in New Issue
Block a user