srgb texture colorspace

Show images with physical colorspace, and srgb colorspace via three methods.

• The top image shows physical colorspace, very non-linear.

• The 2nd shows srgb colorspace, interpolated in srgb (srgb-to-linear happens in shader).

• The 3d shows srgb texture, interpolated in linear space (srgb-to-linear is automatic).

• The 4th is the same, using an explicit texture format.

import numpy as np
from wgpu.gui.auto import WgpuCanvas, run
import pygfx as gfx


renderer = gfx.renderers.WgpuRenderer(WgpuCanvas(size=(600, 600)))

rgba = np.repeat(np.linspace(0, 255, 6).reshape(1, -1), 2, 0).astype(np.uint8)
rgba.shape = (*rgba.shape, 1)
rgba = np.repeat(rgba, 4, 2)
rgba[:, :, 3] = 255
rgb = rgba[:, :, :3].copy()


material = gfx.ImageBasicMaterial(clim=(0, 255), interpolation="linear")

image1 = gfx.Image(
    gfx.Geometry(grid=gfx.Texture(rgb, dim=2, colorspace="physical")),
    material,
)

image2 = gfx.Image(
    gfx.Geometry(grid=gfx.Texture(rgb, dim=2, colorspace="srgb")),
    material,
)
image2.local.y = 2.1

image3 = gfx.Image(
    gfx.Geometry(grid=gfx.Texture(rgb, dim=2, colorspace="tex-srgb")),
    material,
)
image3.local.y = 4.2

# Use explicit format. We still need to set the colorspace, otherwise it converts twice.
# Pygfx will catch and warn for this case.
image4 = gfx.Image(
    gfx.Geometry(
        grid=gfx.Texture(rgba, dim=2, colorspace="tex-srgb", format="rgba8unorm-srgb")
    ),
    material,
)
image4.local.y = 6.3

scene = gfx.Scene()
scene.add(image1, image2, image3, image4)

camera = gfx.OrthographicCamera()
camera.local.scale_y = -1
camera.show_rect(-1, 6, -1, 8)

renderer.request_draw(lambda: renderer.render(scene, camera))


if __name__ == "__main__":
    run()