LlamaEdge-StableDiffusion 0.2.1

Major changes:

• New endpoints
  • GET /v1/files/{file_id}: Retrieve information of a specific file by id
  • GET /v1/files/{file_id}/content: Retrieve the content of a specific file by id
  • GET /v1/files/download/{file_id}: Download a specific file by id

LlamaEdge-StableDiffusion 0.2.0

Major changes:

• (BREAKING) rename --context-type CLI option to --task

• Support control_image field in image generation requests

• Support control_image field in image edit requests

• New CLI options

  • --control-net: Path to control net model. Optional.
  • --control-net-cpu: Keep control net model on cpu. Defaults to false.

• Update the default value of --lora-model-dir CLI option to none

• New document: ENDPOINTS.md

LlamaEdge-StableDiffusion 0.1.7

Major changes:

• Upgrade to llama-core v0.20.0 and endpoints v0.16.0
• Support negative_prompt, cfg_scale, sample_method, steps, height, width, control_strength, seed, and strength params for image-to-image tasks

LlamaEdge-StableDiffusion 0.1.6

Major changes:

• Add --context-type CLI option to allow users to choose which one of text-to-image and image-to-image context or both to create
• Improve the documents

LlamaEdge-StableDiffusion 0.1.5

Major changes:

• Upgrade to llama-core v0.18.1 to create text_to_image and image_to_image contexts during the sd context initialization
• Update docs: README.md, flux.md, and flux_with_lora.md

LlamaEdge-StableDiffusion 0.1.4

Major changes:

• Upgrade to llama-core v0.18.0 to support LoRA models
• Update flux.md
• New example: flux_with_lora.md

LlamaEdge-StableDiffusion 0.1.3

Major change:

• Add --port CLI option

LlamaEdge-StableDiffusion 0.1.2

Major change:

• Migrate to WasmEdge v0.14.1

LlamaEdge-StableDiffusion 0.1.1

Major change:

• New CLI options for standalone diffusion models

  --diffusion-model <DIFFUSION_MODEL>
      Path to the standalone diffusion model file [default: ]
  --vae <VAE>
      Path to vae [default: ]
  --clip-l <CLIP_L>
      Path to the clip-l text encoder [default: ]
  --t5xxl <T5XXL>
      Path to the the t5xxl text encoder [default: ]
  --threads <THREADS>
      Number of threads to use during computation [default: 1]

LlamaEdge-StableDiffusion 0.1.0

Major features:

• /v1/images/generations endpoint for image generation
• /v1/images/edits endpoint for image editing

Models supported:

• second-state/stable-diffusion-v-1-4-GGUF (See README.md for how to run it)
• second-state/stable-diffusion-v1-5-GGUF
• second-state/stable-diffusion-2-1-GGUF
• second-state/FLUX.1-schnell-GGUF (See flux.md for how to run it)