Practical overview
When it fits
Use it when you need a local, repeatable workflow to compare diffusion model outputs or evaluate checkpoints and workflows against selected prompts, seeds and metrics.
Who it is for
- AI researchers comparing diffusion models across prompts, seeds and metrics
- ML engineers evaluating image or video generation quality
- Labs and teams building internal generative-model benchmarking workflows
- Open-source model creators testing checkpoints, LoRAs and workflows
Documented uses
- Compare model outputs across a consistent set of prompts and seeds.
- Evaluate image quality, aesthetics, prompt alignment and object adherence.
- Assess temporal consistency in video outputs and reference-based similarity.
- Keep run context in reproducible logs stored in SQLite.
Requirements & limits
The repository describes DreamLayer Eval as free and open source under GPL-3.0. It is designed to run locally. External model APIs require users to configure their own API keys; local checkpoints must be added separately.
- The README lists Python 3.8+, Node.js 16+ and recommends at least 8 GB of RAM.
- The application uses a React frontend, Flask-based services, SQLite run storage and ComfyUI integration for image workflows.
- The repository is GPL-3.0 licensed.
- The toolkit ships without weights; local models or API credentials are separate inputs.
- The README provides an optional CIFAR-10 dataset download for FID scoring.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install and start DreamLayer Eval locally
Set up the application and open its local frontend.
Before you start
- Python 3.8 or later
- Node.js 16 or later
- 8 GB or more of RAM recommended
- A downloaded copy of the repository
Steps
Run the repository's dependency installation script for Linux, macOS or Windows.
Run the corresponding startup script for your operating system.
Open the frontend at http://localhost:8080.
Expected outcome
The locally started application exposes its frontend at localhost:8080; ComfyUI is listed at localhost:8188.
Guide limitations
- Model weights are not included and must be added separately for local model workflows.
- The README notes that PyTorch setup on macOS may require retries.
Add local checkpoints to the model list
Make locally stored checkpoints and related model files available in the application.
Before you start
- A running DreamLayer Eval installation
- Local .safetensors or .ckpt files
Steps
Download the model files from a source such as Hugging Face, Civitai or your own training runs.
Place full checkpoints in Checkpoints/, LoRA or LoCon files in Lora/, ControlNet files in ControlNet/, and optional VAEs in VAE/.
In the interface, choose Settings and then Refresh Model List.
Expected outcome
The added models appear in the application's dropdowns.
Guide limitations
- DreamLayer Eval does not ship with model weights.
- This procedure covers adding local files; it does not describe API model setup.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 15 October 2026
- Original website title
- TheDesignFounder/DreamLayer-Eval · Official homepage title · retained in its source language