RIOT

RIOT key features:

• open multiple graphic file formats by looking first at the magic number (it does not need file extension to recognize format)including support for uncommon images types (up to 128 BPP, integer and floating point. EX: hdr images, 16 bit grayscale, etc). Adaptive logarithmic tone mapping algorithm (Drago) used for HDR images
• save and optimize JPEG, GIF and PNG with a simple, clean user interface
• works in dual view: (original - optimized image) or single view (optimized image). Automatic preview of resulting image
• in-place compare function (alternativelly display the original image over the optimized image to notice small pixel changes)
• compress files to desired filesize threshold
• fast processing (all is done in memory); see instant results including resulting filesize
• decide if you want to keep metadata (comments, IPTC, Adobe XMP, EXIF profiles, ICC profiles). Unsupported metadata is removed
• transfer metadata between image formats (destination format must support them)
• common tools: pan and zoom, rotate, flip
• resize image by using well known resample filters (ex: Lanczos3, Catmull Rom, Bicubic, and others)
• the compression and the results are comparable to those of commercial products.

Input image types

common bitmap images as well as Adobe Photoshop PSD files.

Optimization details

JPEG settings

You can find the following JPEG settings:

• quality (compression)
• advanced chroma subsampling (none,low 4:2:2,medium 4:2:0,high 4:1:1)
• save as 8bit grayscale
• save progressive

Supported output bit depths: 24 bit color, 8 bit grayscale

GIF settings

The following GIF settings are currently implemented:

• color reduction
  • color quantization downto 256..2 colors (no dithering) using Xiaolin Wu or NeuQuant neural net
  • grayscale 8bit palette
  • floyd-steinberg dithered 1bit monochrome
• save interlaced

Supported output bit depths: 4, 8 bit color, 8 bit grayscale, 1 bit monochrome

PNG settings

The following PNG settings are currently implemented:

• true color support (24 bit RGB, 32 bit RGBA)
• color reduction
  • color quantization downto 256..2 colors (no dithering) using Xiaolin Wu or NeuQuant neural net
  • grayscale 8bit palette
  • floyd-steinberg dithered 1bit monochrome
• save interlaced
• maximum ZLIB compression flag

Automatic pnglib filtering optimizations apply.

Supported output bit depths: 4, 8, 24, 32 bit color, 8 bit grayscale,1 bit monochrome

Metadata settings

Choose to keep or remove the following metadata:

• comments
• Adobe XMP info
• IPTC info
• EXIF profile (including GPS and maker notes)
• ICC color profiles

Unknown or unsupported metadata is automatically removed.

Mask settings

The program lets you choose several transparency options:

• Keep transparency
  (use threshold for conversion from alpha to indexed transparency)
• Blend with solid background
  (choose a color for blending the transparency into the background (alpha composition))
• Set opaque
  (remove transparency information by making all pixels opaque)

Known issues:

• memory leaks when using the IrfanView plugin (especially visible in batch mode) - this will be fixed in the next release

Future plans:

1. batch processing multiple images
2. Adobe Photoshop Filter support (8BF) - add GML Matting plugin for object extraction.
3. twain scanner support
4. basic image processing (brightness, contrast, etc)
5. jpeg smoothening
6. PNG compression optimizations (I need time for this)
7. define transparent regions (with variable transparency)
8. crop function
9. overlay image watermark

You can suggest new features and create polls for implementing them in the official forum.

Available as stand-alone executable (portable app) or dynamic link library for developers. The plugin is already included into the best image viewer worldwide (in my oppinion) - IrfanView. This software uses a modified version of FreeImage open source image library. See http://freeimage.sourceforge.net for details. FreeImage is used under the FIPL license version 1.0.