The following test cases always generate images that differ from run to run:
IMAGE v3d raytrace dof_stereo: dof_stereo.png differs
IMAGE v3d raytrace sample_ball: sample_ball_iss_iss_256.png differs
IMAGE v3d raytrace sample_ball: sample_ball_iss.png differs
IMAGE v3d raytrace sample_ball: sample_ball_iss_iss_64.png differs
IMAGE v3d raytrace sample_ball: sample_ball_iss_iss_1024.png differs
IMAGE v3d raytrace dof_mono: dof_mono.png differs
IMAGE v3d raytrace sample_ball_iss: sample_ball_iss.png differs
IMAGE v3d raytrace sample_ball_iss: sample_ball_iss_iss_256.png differs
IMAGE v3d raytrace sample_ball_iss: sample_ball_iss_iss_64.png differs
IMAGE v3d raytrace sample_ball_iss: sample_ball_iss_iss_1024.png differs
IMAGE v3d raytrace tone_mapping: tone_mapping.png differs
If the algorithm cannot provide the same image on each run then we should avoid generating such images and their comparison. |