Optimize Parameters about pix2model HOT 2 CLOSED

FeatureExtraction
Force CPU Extraction: try without, might be unstable, try with SIFT describer type
Describer types: Here is a survey paper https://research-repository.griffith.edu.au/bitstream/10072/419561/2/Zhang3477189-Accepted.pdf
multiple describer types: dsp-sift + akaze? select the same in the following nodes
Describer Preset: speed-quality tradeoff, try high

ImageMatching
check voctree vs sequential vs exhaustive
only 100 images -> exhaustive ok?
min for voc tree, sequential neighbors; max descriptors 0, Nb matches 0

FeatureMatching
check matching methods: brute force vs heuristics
geometric estimator and filter types
guided matching?
cascade hashing l2; guided matching, cross matching, known poses geometric error max 0

StructureFromMotion
localizer estimator
Observation Constraint (Scaling on)
disable Use Rig Constraint

PrepareDenseScene
disable save meta data?
Correct images exposure?

Depth Map
downscale 1? (longer meshing)

Depth Map Filter
number of nearest cameras 10 -> 15?

SfM Transform
Apply Editor transformations (check if non-uniform scaling is supported)

Meshing
try to get it faster, even if it costs some quality
use custom bounding box from editor
min step 1
margin init (final) coef 1
densify front/back 5
colorize output (vertex colors?)
save raw dense point cloud
max points (per voxel): reduce for speed?
Min Observations Angle For SfM Space Estimation: filter uninteresting remote areas

Mesh Filtering
less smoothing?

Mesh Decimation
target faces <= 600k (LSCM) or 300k (ABF), matching the texturing>unwrap method

Texturing
try to get it faster, even if it costs some quality. prefer smaller files.
texture side and downscale (multi-band?)
Subdiv Target Ratio?
unwrap method: check number of faces, ABF if it doesn't take too long, else LSCM, else basic
correct exposure?

from pix2model.

switching to AliceVision is too much work for now

from pix2model.