Very large image manipulation

What if a bitmap size is much larger than the amount of RAM available?

  • Load and save images of hundreds of MPix
  • Lean memory usage for any sized image
  • Incredible performance
View Demo Download Trial

Processing large images without OutOfMemoryException

The traditional way to process an image is to allocate a buffer large enough to hold all its pixels and decode an image file into memory.

It works great for regular photos, but what if the image dimensions are really large? For example, satellite images? Billboard designs? Panoramic images stitched out of dozens of photos?

The obvious solution is to read an image partially, process one piece of the image, save that part of the image, and then proceed.

Instead of loading a bitmap into memory, Graphics Mill allows you to construct a "pipeline" (graph) of effects and "push" the image through this pipeline. It will get a small piece of input.jpg, rotate it, resize and write it to output.jpg. Then it will proceed with the next chunk of input.jpg until it reaches the end of a file.

Memory friendly does not mean complicated

You may be worried that this chunk-by-chunk processing approach is too difficult and requires writing more code. It really isn't! With a convenient and intuitive API, all the powerful image processing functions of Graphics Mill are seamless and completely automatic.

Moreover, you don't always have to use pipelines. It is absolutely fine to combine both classic and pipeline approaches as best suits your needs.. Code examples and documentation for each feature is available for both classic and pipeline APIs.

Let's compare the classic and pipeline APIs:

Classic API

using (var bitmap = new Bitmap("In.jpg"))           
    bitmap.Transforms.Resize(320, 240);           

Memory-friendly pipeline API

using (var reader = ImageReader.Create("In.jpg"))           
using (var resize = new Resize(320, 240))           
using (var writer = ImageWriter.Create("Out.jpg"))           
    Pipeline.Run(reader + resize + writer);           
Classis API memory consumption.
Pipeline API memory consumption.
View sample

Orders of magnitude faster than other SDKs

When processing thousands of images simultaneously, performance is crucial. We work hard to keep Graphics Mill faster than other well-known imaging SDKs.

Operation Graphics Mill System.Drawing WIC
Resize 10000x10000 to 2000x2000 * 20 ms 199 ms 105 ms
Load 1000x1000 JPEG * 2 ms 23 ms 6 ms
Get 100x100 thumbnail from JPEG 1000x1000 * 5 ms 50 ms 30 ms

*Speeds may vary on your servers due to differences in hardware

We take the algorithms' performance seriously. For example, one of our regression test suites is created especially to ensure that changes in Graphics Mill does not slow it down.

Robust image processing

Graphics Mill is robust enough to handle extremely large image loads. Just feed it with any number of images and Graphics Mill will do a perfect job without defects or failures.