Forum

Author Topic: Photoscan 1.4.5 Sequaio black images  (Read 1787 times)

janberto

  • Newbie
  • *
  • Posts: 5
    • View Profile
Photoscan 1.4.5 Sequaio black images
« on: February 22, 2019, 01:25:10 PM »
Hi,

I am aware of a similar issue at https://www.agisoft.com/forum/index.php?topic=8987.0 . However, there they were complaining about images being black due to the lack of contrast. However, my images are completely black.

When I use Photoscan 1.3.5 (on the left in the attachments) and 1.4.5 (on the right in the attachments) side to side and import literally the same images, the older version of Photoscan shows my the images whilst the new one just shows black images. For both of the version the workflow works. I am able to align the images and create a DEM and Ortho. So the information per image does seem to be there. However, when I export the ortho with the new version, the ortho is completely black. For every single pixel is has a value of 0 for the first 3 band and a value 255 for the last band.

I want to use 1.4.5 because there I can calibrate the reflection.

Does anyone know what I am doing wrong?

Greetings,

Janbert

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14839
    • View Profile
Re: Photoscan 1.4.5 Sequaio black images
« Reply #1 on: February 22, 2019, 01:42:49 PM »
Hello Janbert,

Can you send a small subset of images (just for a few camera locations) to support@agisoft.com, so that we can reproduce the issue?

Also please specify, if you are using Reflectance Calibration operation in 1.4.5; whether you have tried to switch the primary channel to NIR and adjust the brightness and if during the export you are using any Raster Transformation option (like Index Color) - since the source data is four-band 16 bit data, and in the output, according to your description, you have 8-bit RGB + alpha channels.
Best regards,
Alexey Pasumansky,
Agisoft LLC

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14839
    • View Profile
Re: Photoscan 1.4.5 Sequaio black images
« Reply #2 on: February 22, 2019, 06:06:22 PM »
Hello Janbert,

As far as I see from the provided images, the used range in the TIFFs is 0 - 256 values, while the Black Level in the image meta is over 4000 for each band. All the values below this number will be treated as black.

Make sure, that the camera that you are using to capture the images has firmware updated to the latest version.

To "fix" the issue for the given dataset you can manually adjust or discard (set to zero) the black level for each Band in the Camera Calibration dialog.
« Last Edit: February 22, 2019, 06:07:53 PM by Alexey Pasumansky »
Best regards,
Alexey Pasumansky,
Agisoft LLC

janberto

  • Newbie
  • *
  • Posts: 5
    • View Profile
Re: Photoscan 1.4.5 Sequaio black images
« Reply #3 on: February 27, 2019, 01:44:54 PM »
Thank you Alexey for the explanation. I have tried your approach and indeed when I manually set the black level to 0, the image change and visually look the same as when I import them in Photoscan 1.3.5. Then when I process and export the ortho it's not black anymore and it looks fine.

However, when I change the black levels to 0 and afterwards use the 'Calibrate Reflectance' tool to use the multispectral reflectancevalues as input for the Panel Calibration, after the calibration the images turn black again and when processing the whole ortho is also black.

What do you suggest as a solution for this issue?

Greetings,

Janbert

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14839
    • View Profile
Re: Photoscan 1.4.5 Sequaio black images
« Reply #4 on: March 07, 2019, 07:32:19 PM »
Hello Janbert,

Looks like the meta-information related to the radiometric camera model is not correct and doesn't fit actual sensor. I would rather suggest to contact the manufacturer in order to check, if the camera behaves correctly or not.

In case you have the images of the reflectance calibration panel, though, using only panel for the calibration procedure may help to normalize the values properly.
Best regards,
Alexey Pasumansky,
Agisoft LLC