Files
test/tests/python/eevee_next_render_tests.py
Jeroen Bakker eeefd9a69c Vulkan: Enable render tests
This PR enabled backend specific rendertest for EEVEE and Workbench.
Some changes that have been made are:
- Add suffix to the test identifying the backend (_opengl, _vulkan, _metal)
- Vulkan render tests are compared with the opengl results.

Most EEVEE tests run as expected there are some issues in the Vulkan
backend that needs to be addressed:
- Fully smooth reflective materials miss lighting.
- Tangent normals are off

None of the workbench tests pass. It has to do with downloading the depth
buffer. In Workbench they are stored as GPU_DEPTH32F_STENCIL8 and downloaded
as FLOAT. We didn't implement it in the vulkan backend yet and currently asserts.

The Vulkan render test run faster compared to OpenGL. On my system around
25-50% faster.

Pull Request: https://projects.blender.org/blender/blender/pulls/126784
2024-09-05 13:58:14 +02:00

250 lines
7.9 KiB
Python

#!/usr/bin/env python3
# SPDX-FileCopyrightText: 2015-2022 Blender Authors
#
# SPDX-License-Identifier: Apache-2.0
import argparse
import os
import pathlib
import subprocess
import sys
from pathlib import Path
try:
# Render report is not always available and leads to errors in the console logs that can be ignored.
from modules import render_report
class EEVEEReport(render_report.Report):
def __init__(self, title, output_dir, oiiotool, device=None, blocklist=[]):
super().__init__(title, output_dir, oiiotool, device=device, blocklist=blocklist)
self.gpu_backend = device
def _get_render_arguments(self, arguments_cb, filepath, base_output_filepath):
return arguments_cb(filepath, base_output_filepath, gpu_backend=self.device)
except ImportError:
# render_report can only be loaded when running the render tests. It errors when
# this script is run during preparation steps.
pass
# List of .blend files that are known to be failing and are not ready to be
# tested, or that only make sense on some devices. Accepts regular expressions.
BLOCKLIST = [
# Blocked due to point cloud volume differences between platforms (to be fixed).
"points_volume.blend",
# Blocked due to GBuffer encoding of small IOR difference between platforms (to be fixed).
"principled_thinfilm_transmission.blend",
]
def setup():
import bpy
for scene in bpy.data.scenes:
scene.render.engine = 'BLENDER_EEVEE_NEXT'
# Enable Eevee features
eevee = scene.eevee
# Overscan of 50 will doesn't offset the final image, and adds more information for screen based ray tracing.
eevee.use_overscan = True
eevee.overscan_size = 50.0
# Ambient Occlusion Pass
eevee.gtao_distance = 1
# Lights
eevee.light_threshold = 0.001
# Hair
scene.render.hair_type = 'STRIP'
# Shadow
eevee.shadow_step_count = 16
# Volumetric
eevee.volumetric_tile_size = '2'
eevee.volumetric_start = 1.0
eevee.volumetric_end = 50.0
eevee.volumetric_samples = 128
eevee.use_volumetric_shadows = True
eevee.clamp_volume_indirect = 0.0
# Motion Blur
if scene.render.use_motion_blur:
eevee.motion_blur_steps = 10
# Ray-tracing
eevee.use_raytracing = True
eevee.ray_tracing_method = 'SCREEN'
ray_tracing = eevee.ray_tracing_options
ray_tracing.resolution_scale = "1"
ray_tracing.screen_trace_quality = 1.0
ray_tracing.screen_trace_thickness = 1.0
# Light-probes
eevee.gi_cubemap_resolution = '256'
# Only include the plane in probes
for ob in scene.objects:
if ob.type == 'LIGHT':
# Set maximum resolution
ob.data.shadow_maximum_resolution = 0.0
if ob.name != 'Plane' and ob.type != 'LIGHT':
ob.hide_probe_volume = True
ob.hide_probe_sphere = True
ob.hide_probe_plane = True
# Counteract the versioning from legacy EEVEE. Should be changed per file at some point.
for mat_slot in ob.material_slots:
if mat_slot.material:
mat_slot.material.thickness_mode = 'SPHERE'
# Does not work in edit mode
if bpy.context.mode == 'OBJECT':
# Simple probe setup
bpy.ops.object.lightprobe_add(type='SPHERE', location=(0.0, 0.1, 1.0))
cubemap = bpy.context.selected_objects[0]
cubemap.scale = (5.0, 5.0, 2.0)
cubemap.data.falloff = 0.0
cubemap.data.clip_start = 0.8
cubemap.data.influence_distance = 1.2
bpy.ops.object.lightprobe_add(type='VOLUME', location=(0.0, 0.0, 2.0))
grid = bpy.context.selected_objects[0]
grid.scale = (8.0, 4.5, 4.5)
grid.data.resolution_x = 32
grid.data.resolution_y = 16
grid.data.resolution_z = 8
grid.data.bake_samples = 128
grid.data.capture_world = True
grid.data.surfel_density = 100
# Make lighting smoother for most of the case.
grid.data.dilation_threshold = 1.0
bpy.ops.object.lightprobe_cache_bake(subset='ACTIVE')
# When run from inside Blender, render and exit.
try:
import bpy
inside_blender = True
except ImportError:
inside_blender = False
if inside_blender:
try:
setup()
except Exception as e:
print(e)
sys.exit(1)
def get_gpu_device_type(blender):
# TODO: This always fails.
command = [
blender,
"--background",
"--factory-startup",
"--python",
str(pathlib.Path(__file__).parent / "gpu_info.py")
]
try:
completed_process = subprocess.run(command, stdout=subprocess.PIPE)
for line in completed_process.stdout.read_text():
if line.startswith("GPU_DEVICE_TYPE:"):
vendor = line.split(':')[1]
return vendor
except BaseException as e:
return None
return None
def get_arguments(filepath, output_filepath, gpu_backend):
arguments = [
"--background",
"--factory-startup",
"--enable-autoexec",
"--debug-memory",
"--debug-exit-on-error"]
if gpu_backend:
arguments.extend(["--gpu-backend", gpu_backend])
arguments.extend([
filepath,
"-E", "BLENDER_EEVEE_NEXT",
"-P",
os.path.realpath(__file__),
"-o", output_filepath,
"-F", "PNG",
"-f", "1"])
return arguments
def create_argparse():
parser = argparse.ArgumentParser()
parser.add_argument("-blender", nargs="+")
parser.add_argument("-testdir", nargs=1)
parser.add_argument("-outdir", nargs=1)
parser.add_argument("-oiiotool", nargs=1)
parser.add_argument('--batch', default=False, action='store_true')
parser.add_argument('--fail-silently', default=False, action='store_true')
parser.add_argument('--gpu-backend', nargs=1)
return parser
def main():
parser = create_argparse()
args = parser.parse_args()
blender = args.blender[0]
test_dir = args.testdir[0]
oiiotool = args.oiiotool[0]
output_dir = args.outdir[0]
gpu_backend = args.gpu_backend[0]
gpu_device_type = get_gpu_device_type(blender)
reference_override_dir = None
if gpu_device_type == "AMD":
reference_override_dir = "eevee_next_renders/amd"
report = EEVEEReport("Eevee Next", output_dir, oiiotool, device=gpu_backend, blocklist=BLOCKLIST)
if gpu_backend == "vulkan":
report.set_compare_engine('eevee_next', 'opengl')
else:
report.set_compare_engine('cycles', 'CPU')
report.set_pixelated(True)
report.set_reference_dir("eevee_next_renders")
report.set_reference_override_dir(reference_override_dir)
test_dir_name = Path(test_dir).name
if test_dir_name.startswith('image_mapping'):
# Platform dependent border values. To be fixed
report.set_fail_threshold(0.2)
elif test_dir_name.startswith('image'):
report.set_fail_threshold(0.051)
# Noise pattern changes depending on platform. Mostly caused by transparency.
# TODO(fclem): See if we can just increase number of samples per file.
if test_dir_name.startswith('render_layer'):
# shadow pass, rlayer flag
report.set_fail_threshold(0.035)
elif test_dir_name.startswith('hair'):
# hair close up
report.set_fail_threshold(0.0275)
elif test_dir_name.startswith('integrator'):
# shadow all max bounces
report.set_fail_threshold(0.0275)
elif test_dir_name.startswith('pointcloud'):
# points transparent
report.set_fail_threshold(0.06)
ok = report.run(test_dir, blender, get_arguments, batch=args.batch, fail_silently=args.fail_silently)
sys.exit(not ok)
if not inside_blender and __name__ == "__main__":
main()