Approach Faster Animation Playback #73429

Open
opened 2020-01-27 11:13:56 +01:00 by Jeroen Bakker · 22 comments
Member

The Faster Animation Playback second milestone consist out of non-functional requirements. This document will describe the steps we want to follow to execute that milestone.

Goal

The goal of the second milestone of #68908 (Faster Animation Playback) is to increase the playback performance when using Blender as an animator by optimizing the current weaknesses. As viewport rendering and viewport playback are very tight related most changes will impact both of them and all changes should be tested to both use cases.

Approach

Phase 0: Preparation

  • Select several production scenes that will be used as benchmark (@Hjalti)
    ** Scenes from different productions with different rigs and different sizes.
  • Modify blender to profile what is happening during animation playback. (@Jeroen-Bakker
    Depedency Graph: Depedency Graph has a detailed profiler. Just need to check if it is sufficient. Draw Manager: Draw manager already has a detailed profiler. We might want to add small changes (for example separate mesh_extraction from preparation)
    ** Other: We should keep track of the rest of the time (not inside the Dependency graph/Draw manager) to detect balance shifts in performance
  • Create an automated test that records the timings in a database so we could compare the results and make decisions (@Jeroen-Bakker)
  • Document the hardware/OS's for the benchmark that will be used
  • Perform a base benchmark

Phase 1..n: Performance Cycle

When doing performance test you do a single change at a time and monitor what that change does to the whole system. To make a change we suggest the next approach

  • Find the current weakest area
  • Do one or more prototypes to solve the weakest link
  • Select the solution.
  • Create design (mention the other solutions as alternatives) and implement the solution.
  • User test the chosen solution. (@Hjalti)
  • Codereview and merge to master.

Only after we complete a full cycle we can start with the next cycle. We continue this until we reached a workable performance or the year has ended.

Concerns

This approach tends to be too narrow focused when looking for solutions. When looking for solutions we should keep in mind to look at the whole picture and not limit the solution to the area in focus.

Known areas of improvement

This section contains a collection of known issues that we might want to tackle during the project

  • DrawManager mesh extraction uses a task pool per mesh. When two meshes needs to be updated for a frame it happens in serial.
  • Blender's task pool has been setup originally for long running background processes. It is currently used for tasks of all sizes.
  • Better select the frame ranges that needs to be updated for motion curves.
  • VSE/Graph editor dependencies. (ref play in active area)
The Faster Animation Playback second milestone consist out of non-functional requirements. This document will describe the steps we want to follow to execute that milestone. ## Goal The goal of the second milestone of #68908 (Faster Animation Playback) is to increase the playback performance when using Blender as an animator by optimizing the current weaknesses. As viewport rendering and viewport playback are very tight related most changes will impact both of them and all changes should be tested to both use cases. ## Approach ### Phase 0: Preparation * Select several production scenes that will be used as benchmark (@Hjalti) ** Scenes from different productions with different rigs and different sizes. * Modify blender to profile what is happening during animation playback. (@Jeroen-Bakker **Depedency Graph: Depedency Graph has a detailed profiler. Just need to check if it is sufficient.** Draw Manager: Draw manager already has a detailed profiler. We might want to add small changes (for example separate mesh_extraction from preparation) ** Other: We should keep track of the rest of the time (not inside the Dependency graph/Draw manager) to detect balance shifts in performance * Create an automated test that records the timings in a database so we could compare the results and make decisions (@Jeroen-Bakker) * Document the hardware/OS's for the benchmark that will be used * Perform a base benchmark ### Phase 1..n: Performance Cycle When doing performance test you do a single change at a time and monitor what that change does to the whole system. To make a change we suggest the next approach * Find the current weakest area * Do one or more prototypes to solve the weakest link * Select the solution. * Create design (mention the other solutions as alternatives) and implement the solution. * User test the chosen solution. (@Hjalti) * Codereview and merge to master. Only after we complete a full cycle we can start with the next cycle. We continue this until we reached a workable performance or the year has ended. ## Concerns This approach tends to be too narrow focused when looking for solutions. When looking for solutions we should keep in mind to look at the whole picture and not limit the solution to the area in focus. ## Known areas of improvement This section contains a collection of known issues that we might want to tackle during the project * DrawManager mesh extraction uses a task pool per mesh. When two meshes needs to be updated for a frame it happens in serial. * Blender's task pool has been setup originally for long running background processes. It is currently used for tasks of all sizes. * Better select the frame ranges that needs to be updated for motion curves. * VSE/Graph editor dependencies. (ref play in active area)
Author
Member

Changed status from 'Needs Triage' to: 'Confirmed'

Changed status from 'Needs Triage' to: 'Confirmed'
Author
Member

Added subscriber: @Jeroen-Bakker

Added subscriber: @Jeroen-Bakker
Jeroen Bakker changed title from EngineeringPlan: Faster Animation Playback Global Approach to [WIP] Global Approach Faster Animation Playback 2020-01-27 11:53:52 +01:00
Author
Member

Added subscriber: @Hjalti

Added subscriber: @Hjalti

Added subscriber: @MarcelLegindi

Added subscriber: @MarcelLegindi
Jeroen Bakker changed title from [WIP] Global Approach Faster Animation Playback to Approach Faster Animation Playback 2020-01-27 13:17:00 +01:00

Added subscriber: @WilliamReynish

Added subscriber: @WilliamReynish

Here I've often found this rather unexpected culprit that can make playback slow:

Having the Dopesheet vand/or Graph Editor visible can sometime more than halve (!) the playback speed, simply because Blender spends so many resources re-drawing the keyframes and curves, always every frame. This seems silly. Surely it doesn't need to do that, or it could perhaps be optimized?

Here I've often found this rather unexpected culprit that can make playback slow: Having the Dopesheet vand/or Graph Editor visible can sometime more than halve (!) the playback speed, simply because Blender spends so many resources re-drawing the keyframes and curves, always every frame. This seems silly. Surely it doesn't need to do that, or it could perhaps be optimized?

Added subscriber: @Phigon

Added subscriber: @Phigon
Author
Member

@WilliamReynish that is what is referred to with the VSE/Graph editor dependencies. most of the time the whole screen is updated, but the user is just looking at a single area. There seems to be an option that is used by the animators here where only the area is updated where the mouse is Active Editor Only.

@WilliamReynish that is what is referred to with the `VSE/Graph editor dependencies`. most of the time the whole screen is updated, but the user is just looking at a single area. There seems to be an option that is used by the animators here where only the area is updated where the mouse is `Active Editor Only`.
Member

Added subscriber: @EAW

Added subscriber: @EAW

Added subscriber: @Scaredyfish

Added subscriber: @Scaredyfish

Added subscriber: @Coverop

Added subscriber: @Coverop

I would like to add that disabling "Auto smooth" from meshes will increase performance while previewing animation in viewport.

Here's an example:
https://streamable.com/6gkf8

I would like to add that disabling "Auto smooth" from meshes will increase performance while previewing animation in viewport. Here's an example: https://streamable.com/6gkf8

Added subscriber: @0o00o0oo

Added subscriber: @0o00o0oo

Added subscriber: @YutaroFukagawa

Added subscriber: @YutaroFukagawa

Added subscriber: @SamGreen

Added subscriber: @SamGreen

Added subscriber: @pauanyu_blender

Added subscriber: @pauanyu_blender

Added subscriber: @mysticfall

Added subscriber: @mysticfall

Added subscriber: @AquaticNightmare

Added subscriber: @AquaticNightmare
Contributor

Added subscriber: @RedMser

Added subscriber: @RedMser

Added subscriber: @chadking

Added subscriber: @chadking

Added subscriber: @Garek

Added subscriber: @Garek
Contributor

Added subscriber: @Nika-Kutsniashvili

Added subscriber: @Nika-Kutsniashvili
Philipp Oeser removed the
Interest
EEVEE & Viewport
label 2023-02-09 15:15:10 +01:00
Philipp Oeser added the
Interest
Core
label 2023-02-10 11:09:54 +01:00
Sign in to join this conversation.
No Label
Interest
Alembic
Interest
Animation & Rigging
Interest
Asset Browser
Interest
Asset Browser Project Overview
Interest
Audio
Interest
Automated Testing
Interest
Blender Asset Bundle
Interest
BlendFile
Interest
Collada
Interest
Compatibility
Interest
Compositing
Interest
Core
Interest
Cycles
Interest
Dependency Graph
Interest
Development Management
Interest
EEVEE
Interest
EEVEE & Viewport
Interest
Freestyle
Interest
Geometry Nodes
Interest
Grease Pencil
Interest
ID Management
Interest
Images & Movies
Interest
Import Export
Interest
Line Art
Interest
Masking
Interest
Metal
Interest
Modeling
Interest
Modifiers
Interest
Motion Tracking
Interest
Nodes & Physics
Interest
OpenGL
Interest
Overlay
Interest
Overrides
Interest
Performance
Interest
Physics
Interest
Pipeline, Assets & IO
Interest
Platforms, Builds & Tests
Interest
Python API
Interest
Render & Cycles
Interest
Render Pipeline
Interest
Sculpt, Paint & Texture
Interest
Text Editor
Interest
Translations
Interest
Triaging
Interest
Undo
Interest
USD
Interest
User Interface
Interest
UV Editing
Interest
VFX & Video
Interest
Video Sequencer
Interest
Virtual Reality
Interest
Vulkan
Interest
Wayland
Interest
Workbench
Interest: X11
Legacy
Blender 2.8 Project
Legacy
Milestone 1: Basic, Local Asset Browser
Legacy
OpenGL Error
Meta
Good First Issue
Meta
Papercut
Meta
Retrospective
Meta
Security
Module
Animation & Rigging
Module
Core
Module
Development Management
Module
EEVEE & Viewport
Module
Grease Pencil
Module
Modeling
Module
Nodes & Physics
Module
Pipeline, Assets & IO
Module
Platforms, Builds & Tests
Module
Python API
Module
Render & Cycles
Module
Sculpt, Paint & Texture
Module
Triaging
Module
User Interface
Module
VFX & Video
Platform
FreeBSD
Platform
Linux
Platform
macOS
Platform
Windows
Priority
High
Priority
Low
Priority
Normal
Priority
Unbreak Now!
Status
Archived
Status
Confirmed
Status
Duplicate
Status
Needs Info from Developers
Status
Needs Information from User
Status
Needs Triage
Status
Resolved
Type
Bug
Type
Design
Type
Known Issue
Type
Patch
Type
Report
Type
To Do
No Milestone
No project
No Assignees
17 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: blender/blender#73429
No description provided.