Bug report: Incorrect drawing order of Isosurfaces with opacity

Start by running the provided Python reproduction in Chrome and compare the isosurface and volume rendering at opacity 0.99 versus 1. Investigate the rendering path for opaque and nearly opaque Isosurfaces, multisurfaces, and volume plots. Done means the supplied cases draw in the correct order without regressing fully opaque rendering.

• Reported by Łukasz as a post on the forum
• Plotly version: 6.2.0
• Python version: 3.11.13
• Browser: chrome

Hi, i have found quite annoying bug which can be seen only when the isosurface components are opaque. I observed that it doesn’t only occur for multiple trace isosurfaces, but also single multisurface isosurfaces and volume plots.

Here is an example code how to replicate it:
(The fun part is that when you change the opacity from 0.99 to 1 the plot gets visualised properly.)

import plotly.graph_objects as go
import numpy as np
import nrrd
from skimage.transform import resize


volume = np.zeros((50, 50, 50))
volume[1:10, 5:45, 5:45] =1
volume[10:20, 5:45, 5:45] = 1.5
volume[20:30, 5:45, 5:45] = 2
volume[30:40, 5:45, 5:45] = 2.5
volume[40:49, 5:45, 5:45] = 3
Z, X, Y = np.mgrid[0:volume.shape[0]:1, 0:volume.shape[1]:1, 0:volume.shape[2]:1]


fig = go.Figure(data=[go.Isosurface(
    x=X.flatten(),
    y=Y.flatten(),
    z=Z.flatten(),
    opacity=0.99,
    value=np.where(volume == 1, 1, 0).flatten(),
    surface_fill=1,
    surface_count=1,
    colorscale=[[0, 'rgb(0,255,0)'], [1, 'rgb(0,255,0)']],
    caps=dict(x_show=False, y_show=False, z_show=False),
    #slices_z=dict(show=True, locations=list(range(volume.shape[0]))),
    ),
    go.Isosurface(
        x=X.flatten(),
        y=Y.flatten(),
        z=Z.flatten(),
        opacity=0.99,
        value=np.where(volume == 3, 1, 0).flatten(),
        surface_fill=1,
        surface_count=1,
        colorscale=[[0, 'rgb(255,0,0)'], [1, 'rgb(255,0,0)']],
        caps=dict(x_show=False, y_show=False, z_show=False),
        # slices_z=dict(show=True, locations=list(range(volume.shape[0]))),
    )])

The bug can be seen looking form under the plot: