Agisoft PhotoScan 1.4.0 pre-release

[wojtek]

Congrats on the new release.

Can you guys provide some basic rundown on the new features? For instance, is the Refine Mesh command doing procedural detail extraction from the source images (like color to normal software) or is there some logic involved that actually improves upon the original result?

[bisenberger]

Cool!      Impressive list of changes. Look forward to playing with it.

[Kjellis85]

Hello Kjellis85,

Which OS you are using and which codec is used by the video?

i am running Win 10 and the video uses h.264 codec.

[Vlad]

Even if user decimate mesh to a deep **it and will play with 50images, such heado-on solution will always have not enough memory. At least OpenMVS can use CPU and RAM. And 300+ images still possible to process on 64Gb RAM.

Currently refine mesh functionality is experimental feature in PhotoScan and is available on GPU only (CUDA / OpenCL), but we are planning to implement CPU support for this feature in the next updates.

As for the video memory consumption, it depends not on the number of images, but the image resolution and quality option selected in Refine Mesh dialog. Also the mesh size is important, but it's impact is considerably lower than the one related to the images.

Yes i know, i played with this method on OpenMVS, with exactly the same limitations. So CPU version also did not help too much. 64Gb RAM was not enough for 300+ 16Mpx images because all processes going in core.

Do you have any information regarding memory requirements in your implementation?

53x16Mpx images can be processed at High settings, not Ultra.

But what VRAM needed for example for 500x24Mpx on High? Or On lower?

[Mohammed]

Dear Alexey,

Is there any news about improve the speed process many people look for that.

Mohamed

[bisenberger]

Can PhotoScan  1.4.0 pre-release be installed along side PhotoScan 1.3.4?

I think in previous pre-releases it was possible by moving the stable release install folder to a different location.

[engelbeck]

First...Awesome features in the 1.4 Beta release! Imbedding image thumbnails into Camera boxes in Model view really helps visualize which cameras are misaligned!

I think maybe I found a couple of bugs?:

1.) I tried to import a manually created Markers file exported from Photoscan Pro 1.3X into Photoscan Pro 1.4 Beta.  I got a pop-up error dialog saying "Missing Sensor ID"

2.) I tried to import a MP4 video file and got a popup dialog saying "A media resource couldn't be resolved"

[Alexey Pasumansky]

Can PhotoScan  1.4.0 pre-release be installed along side PhotoScan 1.3.4?

I think in previous pre-releases it was possible by moving the stable release install folder to a different location.

Hello Bill,

If you are using Windows, then you need to copy /PhotoScan Pro/ folder to another location before installing 1.4, since Windows will remove the previous installation (like Vlad has mentioned on the first page).

[Alexey Pasumansky]

I think maybe I found a couple of bugs?:

1.) I tried to import a manually created Markers file exported from Photoscan Pro 1.3X into Photoscan Pro 1.4 Beta.  I got a pop-up error dialog saying "Missing Sensor ID"

2.) I tried to import a MP4 video file and got a popup dialog saying "A media resource couldn't be resolved"

Thank you for reporting.

We'll fix the XML compatibility for markers in the next update. As for the video import, please specify the OS you are using and the codec used in the video.

[PolarNick]

Currently refine mesh functionality is experimental feature in PhotoScan and is available on GPU only (CUDA / OpenCL), but we are planning to implement CPU support for this feature in the next updates.

As for the video memory consumption, it depends not on the number of images, but the image resolution and quality option selected in Refine Mesh dialog. Also the mesh size is important, but it's impact is considerably lower than the one related to the images.

Yes i know, i played with this method on OpenMVS, with exactly the same limitations. So CPU version also did not help too much. 64Gb RAM was not enough for 300+ 16Mpx images because all processes going in core.

I am pretty sure that OpenMVS loads all images to VRAM before starting processing, so its video memory consumption depends on the number of images.

[SAV]

Hi Alexey,

Is it possible to generate a custom PATH/TRACK file within PhotoScan (not just horizontal or vertical as listed in the presets) that can be used for animations/fly-through videos?

Would be nice if I could save a number of 'views/viewpoints' and then generate a PATH/TRACK file from it.

Additionally, my 3DConnexcion Mouse only allows me to PAN/MOVE the model/point cloud, but not ROTATE or ZOOM. In the photo view it does not do anything (would be nice if ZOOM & PAN would work here).

Last but not least, what happened to ENABLE COLOR CORRECTION when building a texture or orthomosaic? I couldn't find it.

Regards,
SAV

[Alexey Pasumansky]

Hello SAV,

Currently camera tracks can be imported from .path file that can be created in Blender, or from .fbx file. PhotoScan itself currently doesn't have instruments of manual tracks creation.

Color correction has been moved to the separate procedure: Tools Menu -> Calibrate Colors. The results of color calibration procedure will be applied to the texture/orthomosaic generation stages.

As for the 3Dconnexion mouse, we'll check it shortly.

[SAV]

Hi Alexey,

Thanks for clarifying.

Is there a way to generate a *.path file based on viewpoints using PYTHON scripting? ;-)
Would be great to have the possibility to generate custom fly-through animations without having to use third party software (such as Blender) mid way to generate the path/track file.

Regarding color correction. What does the CALIBRATE WHITE BALANCE do? Shouldn't there be a 'dropper tool' like in Photoshop that would allow you to pick the proper colour for calibration?

One more thing. Is there a way to select the dense point cloud based on image count? This is already possible for the sparse point cloud (gradual selection tool).

Regards,
SAV

[Alexey Pasumansky]

Hello SAV,

There's some limited access to the camera track via:
chunk.animation.track

I've checked 3Dconnexion mouse and haven't got issues with rotation and zooming (the latter is performed by "flying" higher or lower).

For color correction the "white balance" option means that each band will be calibrated individually.

[bisenberger]

Can PhotoScan  1.4.0 pre-release be installed along side PhotoScan 1.3.4?

I think in previous pre-releases it was possible by moving the stable release install folder to a different location.

Hello Bill,

If you are using Windows, then you need to copy /PhotoScan Pro/ folder to another location before installing 1.4, since Windows will remove the previous installation (like Vlad has mentioned on the first page).

Oops! I missed that.
Thanks Alexey