Ok So I don’t know if upping the levels individually to each image would be advantageous in this scenario. Or just pack up and give up We are 1000m into a peaty tunnel (underwater too) and the light gets saturated by the particulate in the water. Thus the cameras cannot pickup enough detail. If I did a batch level change in photoshop and upped the levels would Reality Capture be helped finding points this way or does it not look at a photo this way, and more just about the digital information no matter the level? The detail against a wall is so limited also ad dark that I a not surprised that its struggling. Heres some eg’s.
In a perfect world I would like them to be RAW but sadly they are extract .Jpg’s from two go Pro 3+ Blacks. We initially had the Go pro’s taking stills but found the light level was not enough to see anything at all. So we had to switch to video at 1080p Even then I am now not that hopeful. Reality has regected around 90% of 1900 images and made around 20 odd components with the rest. Basically not good for what we are needing. I just hoped that by simply upping the levels to a good amount in post in photoshop it might allow for reality capture to squeeze out some more info.
No, images are mostly dead. Wall is out of focus and motion blur. So just not possbile recover any significant details with preprocessing. And if stil use some preprocessed, denoised, downscaled images, result will be really poor. Probably only basic shapes.
Hi Thomas
NO, image downscale is primarily used to ( partially ! ) remove the blur in the image. Open any blurred image and downsample it, and you will see it helps get a sharper image ( works just with certain level of blurriness, if it is too high, this will not help )
Thank Wish G am currently beasting through various underwater photo G scan so hopefully this will help me get better results when the pictures are naf.
Hi Thomas
RAW capture is a requirement for low-light scanning. Best if you have a specialised camera like Nikon D810A ( Astro ) or D5 which can capture images with very low noise… as noise is one of the main problems for photogrammetry. look for lower resolution, higher color density cameras…