Silhouettes (in cartoon mode) have a jagged appearance when the scene is rotates

[chongtianyifa]

Hi,
I recently updated to the newest version and found that the rendered brain outline is very coarse, especially during rotating.
The first frame looks ok:

the outline looks coarse after rotating:

media0.mp4

Below is the video created with the same data using an older version (very likely v2.0.0.0）

Media1.mp4

Is there any way to improve the quality?

[alessandrofelder]

Hey @chongtianyifa

Thanks for raising this.
Can you explain how you generate the videos here, please?

[adamltyson]

I've also noticed this, I just haven't had the time to look into it. I think it's specific to the default, cartoon style. Not sure if it's on our side, or on the vedo side.

As an aside, we may want to change the default away from cartoon. It looks nice for 2D figures, but it's less useful for 3D exploration.

[chongtianyifa]

hi @alessandrofelder,
Here is the script I used in Jupyter notebook to create this video:
`from brainrender import Scene, actors
import brainrender
from brainrender.actors import Points, PointsDensity
import vedo
import numpy as np
import pandas as pd
import random
from brainrender.video import VideoMaker

import os
#from itkwidgets import view

vedo.settings.default_backend= 'vtk'
brainrender.settings.BACKGROUND_COLOR = "white"
brainrender.settings.ROOT_ALPHA = 0.1 # transparency of the overall brain model's actor'
brainrender.settings.SHOW_AXES = False
brainrender.settings.SHADER_STYLE = "cartoon"`

scene = Scene(inset=False, atlas_name='allen_mouse_25um', title=video_title) # make sure this is right
scene.add(Points(coordinates, colors="magenta", alpha=0.9, radius=10))
scene.add_brain_region('SSp', alpha=0.25, color="green")

vm = VideoMaker(scene, name=video_name, save_fld=os.path.join(os.path.dirname(ABBA_accum_tranformed_file))) # niters = number of frames
vm.make_video(
azimuth=1, elevation=0, roll=0 #duration=20 fps=15
) # specify how the scene rotates at each frame

scene.close()‘

Indeed, it is the issue of SHADER_STYLE as @adamltyson mentioned . After trying metallic, plastic, shiny, glossy options, they gave me the same video:

B3M3S17_glossy.mp4

The color of each point is not pure magenta while the outline is smooth now.

[adamltyson]

Looking into this further due to https://forum.image.sc/t/brainrender-lines-on-render/96270.

@marcomusy I don't suppose you have any idea why we're seeing these artefacts using one specific style? We didn't see them in previous versions of brainrender, but we had vedo pinned for a long time.

For anyone seeing this, the workaround is to either set the rotation before rendering, or choose another shader style with:

# before creating a scene
from brainrender import settings
settings.SHADER_STYLE = "plastic" # choose from [metallic, plastic, shiny, glossy, cartoon], default is cartoon

[imagesc-bot]

This issue has been mentioned on Image.sc Forum. There might be relevant details there:

https://forum.image.sc/t/brainrender-lines-on-render/96270/3

[adamltyson]

@alessandrofelder shall we make plastic the default style again?

[alessandrofelder]

shall we make plastic the default style again?

Sure (I don't have a strong opinion... I didn't know it was the default at some point)

[marcomusy]

Hi, from a quick look at the code It looks to me that the "cartoon" option of brainrender generates and adds to the scene a "silhouette" object.
For some reason I don't quite understand, the link to the camera is lost (that is why it only looks good on the first frame). I need to investigate that further..
I guess at the moment the best option is to disable it while I figure out.

[adamltyson]

The default has been changed (temporarily) to plastic in #350

[marcomusy]

This minimal example actually works..

from brainrender import Scene
from vedo import settings

settings.default_font = "Theemim"

scene = Scene()

brain = scene.plotter.objects[0]
sil = brain.silhouette().c('blue5').lw(2)

# scene.add_silhouette(sil) # not working
scene.plotter.add(sil)

scene.plotter.show()
# scene.render() # not working

it seems to me that render() does more stuff than simply rendering the scene.

[adamltyson]

Thanks @marcomusy, this line seems to be the culprit.

Leaving camera as the default (None) seems to work fine with our examples. brainrender sets the camera earlier in the call to scene.render() here, so it's not clear to me why the camera needs to passed to plotter.show().

Is it safe to not pass this parameter to plotter.show()?

[marcomusy]

Totally safe :)

[adamltyson]

@marcomusy thanks! I've raised #352 to fix this.

[adamltyson]

Hi @chongtianyifa this issue has now been fixed. If you update brainrender ( pip install brainrender -U) it should work fine.