# ISIS v2 Character Turnaround Sheet Baseline Workflow Graph

Generated: 2026-08-02T16:33:59.196Z Checklist: `MXA-0020` (ISIS2-00218) Asset
Family: `character-turnaround-sheet` Phase: Phase 4 | Milestone: M3 Launch
(2026-07-27 to 2026-10-30)

## Objective

Provide a deterministic baseline text-to-image workflow graph for character
turnaround sheet generation with locked sampler/model defaults and reproducible
orthographic sheet topology.

## Deterministic Defaults

| Setting           | Value                        |
| ----------------- | ---------------------------- |
| `checkpointModel` | `sd_xl_base_1.0.safetensors` |
| `seed`            | `2026072801`                 |
| `width`           | `2048`                       |
| `height`          | `1536`                       |
| `steps`           | `34`                         |
| `cfgScale`        | `6.5`                        |
| `samplerName`     | `dpmpp_2m_sde`               |
| `scheduler`       | `karras`                     |
| `denoise`         | `1`                          |
| `batchSize`       | `1`                          |
| `outputFormat`    | `png`                        |
| `outputChannels`  | `4`                          |

## Input Contract

| Input                 | Type    | Required | Default                                                                                              |
| --------------------- | ------- | -------- | ---------------------------------------------------------------------------------------------------- |
| `character_archetype` | enum    | yes      | `hero`                                                                                               |
| `style_preset`        | enum    | yes      | `stylized`                                                                                           |
| `turnaround_view_set` | enum    | yes      | `front-side-back`                                                                                    |
| `expression_set`      | enum    | yes      | `neutral-plus-3`                                                                                     |
| `turnaround_layout`   | enum    | yes      | `grid-3x2`                                                                                           |
| `negative_prompt`     | string  | no       | `blurry, low resolution, bad anatomy, extra limbs, duplicated face, text, watermark, jpeg artifacts` |
| `seed`                | integer | yes      | `2026072801`                                                                                         |
| `width`               | integer | yes      | `2048`                                                                                               |
| `height`              | integer | yes      | `1536`                                                                                               |
| `steps`               | integer | yes      | `34`                                                                                                 |
| `cfg_scale`           | float   | yes      | `6.5`                                                                                                |

## Graph Nodes

| Node ID | Class                    | Role                  | Output Count |
| ------- | ------------------------ | --------------------- | ------------ |
| `n1`    | `CheckpointLoaderSimple` | model-load            | 3            |
| `n2`    | `CLIPTextEncode`         | positive-conditioning | 1            |
| `n3`    | `CLIPTextEncode`         | negative-conditioning | 1            |
| `n4`    | `EmptyLatentImage`       | latent-init           | 1            |
| `n5`    | `KSampler`               | latent-sample         | 1            |
| `n6`    | `VAEDecode`              | image-decode          | 1            |
| `n7`    | `SaveImage`              | artifact-output       | 0            |

## Graph Edges

| Edge ID | From   | To                |
| ------- | ------ | ----------------- |
| `e1`    | `n1:0` | `n5.model`        |
| `e2`    | `n1:1` | `n2.clip`         |
| `e3`    | `n1:1` | `n3.clip`         |
| `e4`    | `n1:2` | `n6.vae`          |
| `e5`    | `n2:0` | `n5.positive`     |
| `e6`    | `n3:0` | `n5.negative`     |
| `e7`    | `n4:0` | `n5.latent_image` |
| `e8`    | `n5:0` | `n6.samples`      |
| `e9`    | `n6:0` | `n7.images`       |

## Topological Node Order

`n1` -> `n2` -> `n3` -> `n4` -> `n5` -> `n6` -> `n7`

## Research Alignment

| Workflow                           | Path                                                              | Node Classes                                                                                                                                                                               |
| ---------------------------------- | ----------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Base Avatar Generator              | `libs/isis/workflows/src/workflows/avatar/base-avatar.json`       | `CLIPTextEncode`, `CheckpointLoaderSimple`, `EmptyLatentImage`, `FaceDetailer`, `FreeU_V2`, `KSampler`, `LatentUpscaleBy`, `SaveImage`, `UltralyticsDetectorProvider`, `VAEDecode`         |
| Spiritual Teacher Avatar Generator | `libs/isis/workflows/src/workflows/avatar/spiritual-teacher.json` | `CLIPTextEncode`, `CheckpointLoaderSimple`, `EmptyLatentImage`, `FaceDetailer`, `FreeU_V2`, `KSampler`, `LatentUpscaleBy`, `SaveImage`, `UltralyticsDetectorProvider`, `VAEDecode`         |
| Face Swap Avatar Generator         | `libs/isis/workflows/src/workflows/avatar/face-swap.json`         | `CLIPTextEncode`, `CheckpointLoaderSimple`, `ColorMatch`, `ImageUpscaleWithModel`, `KSampler`, `LoadImage`, `ReActorFaceSwap`, `SaveImage`, `UpscaleModelLoader`, `VAEDecode`, `VAEEncode` |

## Acceptance Criteria

- Graph is acyclic and topologically sortable.
- All deterministic defaults are explicitly set (seed, sampler, scheduler,
  checkpoint).
- Node class inventory covers baseline generation path from prompt conditioning
  to persisted image output.
- Output contract aligns to MXA-0019 required output classes.

## Traceability

- Checklist item: `MXA-0020`
- Ticket key: `ISIS2-00218`
- Source references:
  - `docs/domains/isis/remediation-v2/ISIS_REMEDIATION_V2_TICKETS.json`
  - `docs/domains/isis/remediation-v2/ISIS_V2_CHARACTER_TURNAROUND_SHEET_REQUIREMENTS_BRIEF.json`
  - `libs/isis/workflows/src/workflows/avatar/base-avatar.json`
  - `libs/isis/workflows/src/workflows/avatar/spiritual-teacher.json`
  - `libs/isis/workflows/src/workflows/avatar/face-swap.json`
  - `docs/domains/isis/remediation-v2/ISIS_V2_CONTROL_TAXONOMY.json`
  - `docs/domains/isis/remediation-v2/ISIS_V2_OUTPUT_PACKAGE_TAXONOMY.json`
  - `ISIS_TODOS_2.md`
