Description

From UDN:

We are using the Cascade particle system and we want to render the same scene with multiple cameras, so multiple renders are required. In order to ensure that the result of the particle system is consistent every time, we use "seeded"​ instead of "module". The spawn rate has also been changed to constant. We render using the movie render queue.

We did a simple test: render twice with the same camera, but the results are not the same. Attached is a comparison chart of the difference between the two renderings.

​Because of the following problems:

Can a Cascade particle system get deterministic results in multiple renders (including multiple different cameras)? If yes, how to do it. (We can provide engineering documents)
Can Niagara get deterministic results across multiple renders (including multiple different cameras)? If yes, how to do it.

See attached files.

Steps to Reproduce

hint
I set the Emitter's deterministic using seed. And just render the same frame in Sequencer twice to check if every pixel of the Particle rendered is same. 

Finally I found the only Module with Random Tri Coords node caused the difference. I guess other two Random Tri Coord related nodes for the data interface has same problem
 
Rep steps:
Just use MyProject2
1、Open Myproject2; Open World'/Game/laoshi/NewMap.NewMap'; Open LevelSequence'/Game/laoshi/NewLevelSequence.NewLevelSequence'
2、Open Movie Render Queue, add a job to Render certain frame Twice, the 130th frame for an instance as the settings as below
[Link Removed]
3、Compare the two images rendered using Photoshop or whatever image viewer you prefer to see if both images are matching in pixel perfectly

Expectation: Both images should looks exactly the same

Have Comments or More Details?

There's no existing public thread on this issue, so head over to Questions & Answers just mention UE-121453 in the post.

1
Login to Vote

Won't Fix
ComponentUE - Niagara
Affects Versions4.26.14.27
Target Fix5.4
CreatedAug 11, 2021
ResolvedFeb 12, 2024
UpdatedApr 3, 2024
View Jira Issue