Jump to content

PTM blown out


Recommended Posts

I am in the middle of building an RTI. The object has a very shallow relief inscription. We took a total of 77 images of various angles of raking light from lower degrees of angle in order to view the inscription (5-50 degrees). When I build the ptm, the final result, with the viewing in the middle, is 'blown out.' I did try to rebuild with 51 images -- I pulled out images that seemed too low of raking light and perhaps too high. My results were worse! After further inspection, I did notice that the spheres are reflecting into each other--perhaps they were too close? Any advice?

Link to comment
Share on other sites

Dear T,


I will attempt to be of assistance to you.


The object.

Can you please tell me what kind of material(s) you're shooting. A brief description about its physical characteristics too, to assist in trouble shooting? thanks.


77 Input images.

77 imput images seems like a fairly high number. While its can be good to over shoot an object to get lots of data, too many images can contribute to an abundance of very similar light positions. We recommend that about 36-48 images for your input data set. If you have 77 --- I'd recommend scaling down. And if your images are blown out, definitely weed out the 'hot' overly white looking images. These would typically be the images shot at the high degree angles (the 40-65 degree angles). Think 'Hot images with out decent pixel data' --- dont process them.


Post Processing & EV

If all of your images are running 'hot'. In post processing you can knock back the EV (exposure value) when you are in ACR. Bringing down that EV will make your images darker. (you know this, but just offering a gentle reminder). IF you do this, modify the *batch* -- all the images. (never modify singles, or be tempted to modify the hot images). Think Batch process.


HSH Format.

You also might want to consider re-processing your data as a HSH format. Geek name = Hemispherical Harmonics. In basic description, the HSH rti format uses a different mathematical algorithm in its formula to build the final product. Typically this HSH format does better with very shinny materials and objects that have lots of shadow issues. Depending on the type of material being examined, you might get extra ordinary results in the final RTI. Couple things to keep in mind: the HSH file is considerably larger, and there are not as many viewing modes.


In summary: don't process the hot images, reduce the number of input files, and try the HSH format.


hope that this helps.


Link to comment
Share on other sites

[attachment=0]SphereMap.JPG[/attachment]Hi Marlin,


Thank you for the detailed information! We are shooting marble. The surface is quite dull--non reflective. The inscription is very shallow.


I think the issue of 'overshooting' of same/similar angles with overlapping exposure may be the key to the 'blowing out.' I'll have to re-evaluate the batch and pull "duplicate" angles of light. The highlight composite of the sphere looks good, but I do see some overlap. I've uploaded the image as an attachment for your reference.


All the capture images have perfect density. There are no over-exposed images in the batch. There are, however, a handful of images that are at very low raking light that go into deep shadow. I did remove those images, as well as images that were shot at a higher angle that filled the object, with no luck. This is why I think the overlap of light may be the key.


I tried to build as HSH, but at the final screen the HSH fitter was not loading on it's own. I tried to go find it, but I did not know where to get it. I thought it was part of the software (built-in) when you choose the HSH option - unlike the PTM fitter, right? Does that make sense? Perhaps I need to reload the software?


What about the spheres being too close? Is that an issue?


Thanks, again, for all your help. I can't wait to see the final results on this. It's going to be so cool!



Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...