document some new variables and endpoints

docs/api.md

@@ -59,27 +59,27 @@ Introspection endpoint.
 
 #### `GET /api/settings/filters`
 
-TODO
+List available mask and source filters.
 
 #### `GET /api/settings/masks`
 
-Available mask filters.
+List available mask filters.
 
 #### `GET /api/settings/models`
 
-Available models, all types.
+List available models, all types.
 
 #### `GET /api/settings/noises`
 
-Available noise sources.
+List available noise sources.
 
 #### `GET /api/settings/params`
 
 Server parameters, slider min/max and version check.
 
 #### `GET /api/settings/pipelines`
 
-TODO
+List available pipelines.
 
 #### `GET /api/settings/platforms`
 
@@ -91,11 +91,11 @@ Available pipeline schedulers.
 
 #### `GET /api/settings/strings`
 
-TODO
+Server strings, from extras file and plugins.
 
 #### `GET /api/settings/wildcards`
 
-TODO
+List available wildcard paths.
 
 ### Pipelines
 
@@ -105,11 +105,11 @@ Check if a pipeline has completed.
 
 #### `POST /api/blend`
 
-TODO
+Blend two images using a mask.
 
 #### `POST /api/chain`
 
-TODO
+Run a [custom chain pipeline](./chain-pipelines.md).
 
 #### `POST /api/img2img`
 
@@ -125,15 +125,17 @@ Run a txt2img pipeline.
 
 #### `POST /api/txt2txt`
 
-TODO
+**Unstable.**
+
+Run a txt2txt pipeline.
 
 #### `POST /api/upscale`
 
-TODO
+Run an upscale pipeline.
 
 #### `PUT /api/cancel`
 
-TODO
+Cancel a pending or in-progress job.
 
 ### Outputs
 

docs/chain-pipelines.md

@@ -94,33 +94,33 @@ in [`common/pipelines`](../common/pipelines).
 
 #### Blend: Denoise
 
-TODO
+Run [fast non-local means denoising](https://docs.opencv.org/4.8.0/d5/d69/tutorial_py_non_local_means.html) using `cv2`.
 
 #### Blend: Grid
 
-TODO
+Combine the source images into a grid.
 
 #### Blend: Img2img
 
-TODO
+Run an img2img pipeline.
 
 #### Blend: Linear
 
-TODO
+Blend two images using linear interpolation (0.0 is the first image, 1.0 is the second).
 
 #### Blend: Mask
 
-TODO
+Blend two images using a mask.
 
 ### Correction Stages
 
 #### Correct: CodeFormer
 
-TODO
+Run correction using CodeFormer.
 
 #### Correct: GFPGAN
 
-TODO
+Run correction using GFPGAN.
 
 ### Compound Stages
 
@@ -138,11 +138,11 @@ TODO
 
 #### Persist: Disk
 
-TODO
+Save all of the sources to disk.
 
 #### Persist: S3
 
-TODO
+Save all of the sources to an S3 bucket.
 
 ### Reduction Stages
 
@@ -158,29 +158,29 @@ TODO
 
 #### Source: Noise
 
-TODO
+Create a new source using a noise generator.
 
 #### Source: S3
 
-TODO
+Load a new source from an S3 bucket.
 
 #### Source: Txt2img
 
-TODO
+Run a txt2img pipeline.
 
 #### Source: URL
 
-TODO
+Load a new source from a URL.
 
 ### Upscaling Stages
 
 #### Upscale: BSRGAN
 
-TODO
+Upscaling stage using BSRGAN.
 
 #### Upscale: Highres
 
-TODO
+Upscaling stage using highres.
 
 #### Upscale: Outpaint
 
@@ -195,7 +195,7 @@ Upscaling stage using the Real ESRGAN upscaling models, available in x2 and x4 v
 
 #### Upscale: Simple
 
-TODO
+Upscaling stage using bilinear or Lanczos upscaling.
 
 #### Upscale: Stable Diffusion
 
@@ -206,4 +206,4 @@ Upscaling stage using the Stable Diffusion x4 upscaling model:
 
 #### Upscale: SwinIR
 
-TODO
+Upscaling stage using SwinIR.

docs/server-admin.md

@@ -206,26 +206,32 @@ These extra images can be helpful when debugging inpainting, especially poorly b
 #### Client Variables
 
 - `ONNX_WEB_CIVITAI_ROOT`
-  - TODO
+  - root URL for downloading Civitai models
+  - rarely needs to be changed
 - `ONNX_WEB_CIVITAI_TOKEN`
-  - TODO
+  - Civitai API token for models that require login
+  - you can create an API token in the Profile page: TODO LINK
 - `ONNX_WEB_HUGGINGFACE_TOKEN`
-  - TODO
+  - Huggingface API token for models that require login
 
 #### Conversion Variables
 
 - `ONNX_WEB_CONVERT_CONTROL`
-  - TODO
+  - convert ControlNet
+  - disable to skip ControlNet, saving some disk and memory
 - `ONNX_WEB_CONVERT_EXTRACT`
-  - TODO
+  - extract models to Torch directory before converting to ONNX
+  - disable to skip extraction, saving some disk
+  - extraction uses an older code path that does not work with some newer models and has a non-commercial license
 - `ONNX_WEB_CONVERT_RELOAD`
   - TODO
 - `ONNX_WEB_CONVERT_SHARE_UNET`
   - TODO
 - `ONNX_WEB_CONVERT_OPSET`
   - TODO
 - `ONNX_WEB_CONVERT_CPU_ONLY`
-  - TODO
+  - perform conversion on the CPU, even if a CUDA GPU is available
+  - can allow conversion of models that do not fit in VRAM
 
 #### Memory & Optimization Variables
 
@@ -268,20 +274,22 @@ These extra images can be helpful when debugging inpainting, especially poorly b
 #### Server & Other Variables
 
 - `ONNX_WEB_ADMIN_TOKEN`
-  - TODO
+  - token for admin operations
+  - required to update extras file and restart workers
 - `ONNX_WEB_CORS_ORIGIN`
   - comma-delimited list of allowed origins for CORS headers
 - `ONNX_WEB_DEBUG`
-  - TODO
+  - wait for debugger to be attached before starting server
 - `ONNX_WEB_EXTRA_ARGS`
   - extra arguments to the launch script
   - set this to `--half` to convert models to fp16
 - `ONNX_WEB_FEATURE_FLAGS`
-  - TODO
+  - enable some [feature flags](#feature-flags)
 - `ONNX_WEB_IMAGE_FORMAT`
   - TODO
 - `ONNX_WEB_JOB_LIMIT`
-  - TODO
+  - number of jobs to run before restarting workers
+  - can help prevent memory leaks
 - `ONNX_WEB_PLUGINS`
   - TODO
 - `ONNX_WEB_SERVER_VERSION`
@@ -290,7 +298,8 @@ These extra images can be helpful when debugging inpainting, especially poorly b
   - show progress bars in the logs
   - disabling this can reduce noise in server logs, especially when logging to a file
 - `ONNX_WEB_WORKER_RETRIES`
-  - TODO
+  - number of times to retry a tile or stage before failing the image
+  - more retries takes longer but can help prevent intermittent OOM errors from ruining long pipelines
 
 ### Feature Flags