Need Help Working with RAW files for RC workflow

So I’m just looking for some post processing advice here, I want to use RAW for the texturing and clarity, but not being familiar with the workflow before using RC I seem to be doing it all wrong.

I’m using a camera from the sony a6000 line, great cameras, but aren’t considered the best for JPGs. Some of that however, is because rival cameras put a bit more saturation and kick into their JPGs while Sony’s are a little more true to life (and therefore flatter). Their RAW capability is supposed to be very impressive.

I get more alignments with the JPGs from my shoot than I can from the same RAW files. So I’m not doing something right in Light Room (or Capture One, which I’ve been also using - I don’t want to learn another program in this area unless my life depends on it). I’m losing about 20 percent of processed tiffs that find alignment when I compare it to JPGs in RC. Sometimes I can connect a few components with control points, but for some shots it’s been a losing battle and I revert back to JPGs and the crappier textures.

So what am I doing wrong? Lightroom (LR) / Capture One (C1) have solid camera and lens profiles that do the basics like removing distortion and vignetting without any worries, and the 14-bit files let me almost completely remove highlights and pull shadows with no noticeable difference in image quality. On the other hand, while it looks a lot better when I pixel peep, I don’t know if I’m removing too many highlights or pulling too much shadow - does it ruin RC’s ability to find tie points?

Generally I correct the obvious (distortion, etc.), tweak highlights and shadows and exposure, export to 16 bit TIFF and am always disappointed at the RC result. I’ve tried messing with other things, such as clarity and color and have only made it worse (but I only slightly understand what I am doing in these areas so lay off of it).

I can’t find good tutorials as they are all talking about processing RAW files for portraits or other fanciness that have nothing to do with improving photogrammetry and all about improving artistic photography.

A lot of my work lately has been indoor shooting, so my exposure is all over the place as I jam myself in a corner between four objects and horrible lighting to get an object with good detail. Again, RC manages to stitch all of this with the JPGs, but not nearly as much so as the RAWs. Should I be tweaking exposure for each shot as some go from 1/15th of a second to 4 seconds? I’m taking thousands of pictures so REALLY want to avoid that.

Any advice would be welcome - all the other stuff in my workflow I’ve been able to figure out with research, but all the post processing stuff seems so far away from photogrammetry I don’t know my next steps on how to improve this and I think my work will get a lot better/faster when I figure it out.

Thanks!

I would suspect that the problem is removing the lens distortion. By removing it using software you are not perfectly removing so you end up w/ a subtle (less obvious to the human eye) but more complex lens distortion that is harder to solve and no longer fits the lens model distortion that RC is expecting.

Greg,

Thanks! I’ve messed with the distortion a bit and ran other reconstructions as a result but have gotten the same results… I don’t think distortion is the made issue as the lens and camera profiles in C1 and LR are pretty good at sorting that. I know from my research that other people get better results with RAW files but I’m still doing something wrong in the workflow. I think it may have something to do with exposure, highlights, shadows, and contrast.

Anyone have any other insight for processing RAW files for RC?

I’m thinking now that the messing with the camera database settings in RC may fix my problems. I know I can get better scans with my RAWS.

Can someone tell me if the camera database has a profile for the Sony a6300? I have an a6500 that I’ll copy most of, find the distortion profiles on my lens and dial them in and see if that helps.

Castlenock wrote:

I’m thinking now that the messing with the camera database settings in RC may fix my problems. I know I can get better scans with my RAWS.

Can someone tell me if the camera database has a profile for the Sony a6300? I have an a6500 that I’ll copy most of, find the distortion profiles on my lens and dial them in and see if that helps.

Hello again,
when you search for sensorsdb.xml in your Windows search, you can check it for yourself.
See also: Yellow triangle with an exclamation mark

Thanks lubenko! Will research it today and start experimenting with the next render. Appreciate the response, sure ya’ll are swamped.

As far as I can tell, RC still does not support anything higher than 8bit color space, so it is converting all RAW files to 8bit internally before continuing to alignment. Your camera most likely does a better job at extracting matching details when converting to JPG.

Thanks!

I’ve started a new project and have a lot more experience with my RAW hits - I went out and hit lynda.com for some tutorials and made more thoughtful manipulations before hitting the export button which has resulted in much, much better acquisition

For this scan I’ve been super happy with the initial alignment but it’s 24+ hrs. on 1.8k shots for a high resolution recon (with maddeningly frequent crashes right before finishing). The pictures I found out are accidentally 16 bit TIFF, not 8 bit. RC doesn’t seem to want any more ram than maybe 60 percent of what I have…

Should I stop another 4 hours in on my 5th attempt (changing something each time of course), spend 3 hours having LR chew out new tiffs and start over on the recon? Would that make a difference? I had long recons before for similarlarly complex scenes but not THIS long.

My computer specs: Intel 4.0 Ghz 6700K (stepped down from 4.6 Ghz long ago just in case the OC was doing something weird)
32 GB of RAM
2 x Nvidia 980 ti
All recon files, including images, are off a m.2 array with ‘3GB R/W speeds’. I put that in quotes as I suspect the program recording the value got confused. Still, super-fast.

Hi Castlenock,

I hope this post reaches you with the weird “new” forum…

Did you get more insights with RAW processing or rather directly using them in RC?

I thought from my few tries that RC seems to use the integrated JPGs instead of the RAW data.

Would you be interested in an exchange outside the forum for that topic?