which operations require GPU?

[Dani_rm (ModellingAir)]

Hi,
currently we are running RC on our instances with good CPUs and GPUs, but that’s too expensive.
We were thinking of doing all operations that don’t require GPU on CPU machines only.
Can you please tell me which parts of modelling lifecycle doesn’t require GPU?
Our current steps:

1. load images
2. align
3. build BB
4. reconstruct
5. texture
6. export

Thank you in advance!

[OndrejTrhan (otrhan)]

Hi Thomas,
there you can see the HW usage in each process step:

Alignment:
Feature detection (CPU, RAM, Storage only)
Matching (CPU, RAM, Storage only)
Aligning Images (CPU, RAM, Storage only)
Processing (CPU, RAM, Storage only)

Reconstruction:
Depth map calculation (Storage, RAM, GPU - CUDA, Vram only)
Decompose for Large scene reconstruction
Depth map Grouping (CPU, RAM, Storage only)
Model creation / Meshing (CPU, RAM, Storage only)
Meshing (CPU, RAM, Storage only)

UV unwrap (CPU, Storage, RAM only)

Texture reprojection (CPU, Storage, RAM only)

Texturing:
Model preprocessing (Storage, RAM, light CPU, GPU - CUDA, Vram only)
Model Texturing (heavy CPU, Storage, RAM only)

[Dani_rm (ModellingAir)]

Thank you very much. Can you please align your answer with reality capture cli commands?
F.e.
for aligning we just use:

1. align
2. selectMaximalComponent

for reconstruct:

1. calculateNormalModel
2. selectMaximalComponent
3. exportReport
4. simplify

for texturize:

1. unwrap
2. exportReport
3. calculateTexture

And then export:

1. exportModel

[OndrejTrhan (otrhan)]

Hi Thomas,
basically all relevant processes are described in my previous answer.
So,
-align is relevant to Alignment part
-calculateNormalModel uses Reconstruction part
-simplify should also use Reconstruction part, but probably just Model creation
-unwrap is for UV unwrap
-calculateTexture Texturing

[Dani_rm (ModellingAir)]

thank you