Document duration-based Continuum runtime estimates
This commit is contained in:
@@ -112,6 +112,27 @@ count at two and increases `chunk_seconds` from 6 through 10. The bypass
|
|||||||
counterparts use the same settings. This shows whether longer individual
|
counterparts use the same settings. This shows whether longer individual
|
||||||
clips are viable before combining them into 30-second projects.
|
clips are viable before combining them into 30-second projects.
|
||||||
|
|
||||||
|
## Runtime planning
|
||||||
|
|
||||||
|
The latest observed reference is **18 minutes for one 4-second chunk**. Until
|
||||||
|
more measurements are available, the monitoring estimate is:
|
||||||
|
|
||||||
|
`expected minutes = 18 × (total generated seconds ÷ 4) + 5-minute buffer`
|
||||||
|
|
||||||
|
Examples:
|
||||||
|
|
||||||
|
| Generated duration | Estimated render | Monitoring window |
|
||||||
|
|---:|---:|---:|
|
||||||
|
| 4s | 18m | 23m |
|
||||||
|
| 5s | 22.5m | 28m |
|
||||||
|
| 8s | 36m | 41m |
|
||||||
|
| 20s | 90m | 95m |
|
||||||
|
| 30s | 135m | 140m |
|
||||||
|
|
||||||
|
These are planning windows, not performance guarantees. Each completed test
|
||||||
|
must replace its estimate with the measured runtime and the model should be
|
||||||
|
recalibrated if multi-clip scaling is not approximately linear.
|
||||||
|
|
||||||
The bypass series uses the identical seed, references, prompt text, model,
|
The bypass series uses the identical seed, references, prompt text, model,
|
||||||
samplers, scheduler, upscale settings, tile geometry, and chunk plan. Only the
|
samplers, scheduler, upscale settings, tile geometry, and chunk plan. Only the
|
||||||
Spectrum path changes. Results should be compared in pairs: CT-01 vs CT-B01,
|
Spectrum path changes. Results should be compared in pairs: CT-01 vs CT-B01,
|
||||||
|
|||||||
@@ -13,7 +13,11 @@
|
|||||||
"upscale_denoise": 0.2,
|
"upscale_denoise": 0.2,
|
||||||
"upscale_cfg": 1.0,
|
"upscale_cfg": 1.0,
|
||||||
"tile": "1056x640 with 96px overlap and 32px fade",
|
"tile": "1056x640 with 96px overlap and 32px fade",
|
||||||
"references": ["character-reference.png", "scene-reference.png"]
|
"references": ["character-reference.png", "scene-reference.png"],
|
||||||
|
"runtime_estimate": {
|
||||||
|
"observed_single_chunk": "4 seconds took 18 minutes",
|
||||||
|
"formula": "18 * (total_generated_seconds / 4) + 5 buffer minutes"
|
||||||
|
}
|
||||||
},
|
},
|
||||||
"tests": [
|
"tests": [
|
||||||
{"id":"CT-01","purpose":"2-chunk compact prompt smoke test","prompt_variant":"compact","chunks":2,"chunk_seconds":4,"target_duration_seconds":8,"workflow":"workflow.json","status":"running","prompt_id":"9c31d895-8ccf-4dd2-aefc-da13938f7418"},
|
{"id":"CT-01","purpose":"2-chunk compact prompt smoke test","prompt_variant":"compact","chunks":2,"chunk_seconds":4,"target_duration_seconds":8,"workflow":"workflow.json","status":"running","prompt_id":"9c31d895-8ccf-4dd2-aefc-da13938f7418"},
|
||||||
|
|||||||
@@ -25,5 +25,12 @@
|
|||||||
"latent_upscaler": "minimax_h3_latent_upscaler_3d_conv_v1_fp16.safetensors"
|
"latent_upscaler": "minimax_h3_latent_upscaler_3d_conv_v1_fp16.safetensors"
|
||||||
},
|
},
|
||||||
"allowed_test_variables": ["prompt_variant", "chunks", "chunk_seconds"],
|
"allowed_test_variables": ["prompt_variant", "chunks", "chunk_seconds"],
|
||||||
|
"runtime_estimate": {
|
||||||
|
"observed_single_chunk_seconds": 4,
|
||||||
|
"observed_single_chunk_minutes": 18,
|
||||||
|
"safety_buffer_minutes": 5,
|
||||||
|
"formula": "18 * (total_generated_seconds / 4) + 5",
|
||||||
|
"note": "Planning estimate only; record actual runtime for every result and recalibrate if scaling is not linear."
|
||||||
|
},
|
||||||
"target_duration_seconds": 20
|
"target_duration_seconds": 20
|
||||||
}
|
}
|
||||||
|
|||||||
Reference in New Issue
Block a user