Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
37 changes: 34 additions & 3 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -30,9 +30,10 @@ We believe openness drives innovation, and we invite the research community to i
3. [Performance](#performance)
4. [Quick Start](#quick-start)
5. [Model Summary](#model-summary)
6. [Prompting Guide](#prompting-guide)
7. [Documentation](#documentation)
8. [Citation](#citation)
6. [Ideogram 4 vs Krea 2](#ideogram-4-vs-krea-2)
7. [Prompting Guide](#prompting-guide)
8. [Documentation](#documentation)
9. [Citation](#citation)

## News

Expand Down Expand Up @@ -261,6 +262,34 @@ For full architecture details, see
how the pipeline components fit together, see
[docs/pipeline.md](docs/pipeline.md).

## Ideogram 4 vs Krea 2

Ideogram 4 and Krea 2 are both modern latent flow / rectified-flow image models
built around Diffusion Transformer backbones and Qwen3-VL-class conditioning.
The practical difference is where each system spends its complexity budget:
Ideogram 4 exposes a concrete design-control inference stack in this repo, while
Krea 2's technical report emphasizes broad aesthetic exploration, style
references, prompt-expander training, and a large multi-stage post-training
pipeline.

```mermaid
flowchart LR
Shared["Shared modern recipe: DiT + flow + VLM conditioning"] --> I["Ideogram 4"]
Shared --> K["Krea 2"]
I --> IC["Structured JSON captions"]
I --> IL["Bounding-box layout"]
I --> IT["Typography and palette control"]
I --> IA["Asymmetric CFG inference"]
K --> KD["Data curation pipeline"]
K --> KP["Prompt expander"]
K --> KS["Style references"]
K --> KR["SFT / PO / RL stack"]
```

See [docs/krea2_comparison.md](docs/krea2_comparison.md) for the full
side-by-side comparison and [docs/diagrams.md](docs/diagrams.md) for the diagram
gallery.

## Prompting Guide

Ideogram 4 is trained exclusively on **structured JSON captions**. While
Expand Down Expand Up @@ -312,6 +341,8 @@ See [docs/prompting.md](docs/prompting.md) for a full guide.
| [docs/inference.md](docs/inference.md) | Sampler presets, parameter reference, resolutions, optimization tips |
| [docs/model_architecture.md](docs/model_architecture.md) | Architecture diagram, DiT spec, component details |
| [docs/pipeline.md](docs/pipeline.md) | Conceptual pipeline walkthrough — how all components fit together |
| [docs/krea2_comparison.md](docs/krea2_comparison.md) | Side-by-side comparison of Ideogram 4 and Krea 2 |
| [docs/diagrams.md](docs/diagrams.md) | Mermaid diagrams for the repo, inference flow, CFG, and model comparison |
| [docs/development.md](docs/development.md) | Dev setup, pre-commit hooks, contributing |
| [docs/safety.md](docs/safety.md) | Pre-training, post-training, and inference-time safety mitigations; how to report violations |

Expand Down
Loading