Repository navigation
Stereoscopic view for first ship (anaglyph) #110
Description
Activity
Sources of stereoscopic images for testing:
https://filesamples.com/formats/jps
https://github.com/Bill-Costa/3D-Test-ImagesSupporting stereo 3D add some complexity that might trigger architectural reorganization.
- The logical size of the image is usually diferent than the physical size. This is also true of images with EXIF orientation metadata, but it might be possible to have both situations at once. So we might want a new coordinate frame name.
- There are usually two pixel under the cursor. That's also true of projected dual fisheye images, but in that case it's easy to choose a "dominant" pixel to report. For stereo there are two equally valid pixels. Perhaps the status bar color widget could be doubled? Or toggle between left dominant and right dominant?
- We'll need a new class of shaders. This might motivate a rearrangement of the shader system.
- What happens when we zoom in to individual pixels? Jumbling the two images might be confusing. Actually, if you are viewing correctly, each eye sees a different pixel.
Initial pilot:
- JPS extension is the metadata
- Left and Right images are assumed to be the same size
Why
Stereoscopic viewing is philosophically core to PixelSpy's mission — images as data, the viewer that doesn't lie to you — but nothing stereo has shipped yet. A minimal stereo view in v1 makes the mission honest instead of aspirational.
v1 scope
Blessed second option (post-v1)
Row-interlaced output for passive polarized 3D monitors (e.g. Zalman). The natural next mode, but not v1.
Explicitly out of scope for v1
The rest of the stereo view collection (side-by-side, quad-buffer, etc.). They can follow once v1 ships — the 3-month Store reservation clock is the discipline here.
Not this issue
Distinct from #53 (VR headset view): no headset, no VR thread, no OpenXR handshake. The #53 feature toggle (#108) is unaffected.
Filed by cashew-hamper-bot at Christopher's request.