Jump to content

RTI and photographic materials


mjm

Recommended Posts

Hello,

 

Since I work with photographics materials and the light control concept making RTI possible is essential when making ID and diagnostics in any kind of surfaces, I started making RTI from photo processes like ferrotypes, daguerreotypes (with and without cases) and another.

 

I found that it can be also a great way for studing fungus progression in some surfaces.

 

I will like to discuss with people having working with this kind of objects and RTI in order to learn from their experiences and knowing theirs points of view in several subjects.

 

Anybody here?

 

Kind regards,

 

MJM

Link to comment
Share on other sites

The imaging lab at the Museum Conservation Institute at Smithsonian has posted a few "guest blogs" on the CHI blog.  This one mentions daguerrotypes, though there isn't a lot of information. 

 

We have shot a number of different kinds of photographs including dags. These are mostly as part of training or consulting sessions, so we aren't the experts on what folks are looking for, and how they ultimately ended up using RTI in this work.

 

Carla

Link to comment
Share on other sites

Thanks for these quick and cool answers.

 

I let an small video from a rti file I just made some days ago in order to get some advice on it. There's some things I think are quick bad, and I will like to know also your opinion in including shots that are very reflective "burn outs"but that are inevitable because the reflective surfaces and that shows in some way some interesting information in other parts of the image, for example. Also, in this case I see a difference or blur between the images with lateral light from the left against the ones from the right, but am positive that the camera didn't move or so... Anyways, any feedback will be highly appreciated. (the reaction of the RTI Viewer is slow because my computer is slow).

 

 

am sorry but I don't know how to insert youtube videos here yet...

Link to comment
Share on other sites

Hi,

 

In addition to the links provided by Carla and leszekp, there's a Smithsonian technical note that suggests using a snoot on the light source to reduce the flare on daguerreotypes:  http://www.si.edu/content/MCIImagingStudio/papers/FlexibleSolutions_RTI_Technical_Note.pdf

 

Assuming you don't want to re-capture the images with a modified light source, there are a couple of alternatives you can try using the existing data set.  As you mentioned, there's a trade-off when deciding whether to exclude images with reflections or other localized problems, which still contain useful information.  You can re-process the RTI with and without the problematic images, and refer to different files depending on which details are most important to you.  However, this has the disadvantages of requiring more storage and a means of remembering which versions of the RTIs you want to use for a particular purpose, and complicates archiving.  You can also try processing the images using the HSH fitter (.rti format), which helps with some reflective objects.

 

The default view doesn't seem over-exposed, except maybe the lower left of the subject at high light angles.  The blurriness or ghosting that you see in specular enhancement (at around 1:45 - 2:00 minutes in your video) is odd, since you said you're sure there was no camera movement.  I could guess that it might be from the reflections that occurred at some light angles, or is it possible that the spheres moved during the capture?  This can be checked during the processing.  It almost looks like a self-shadowing effect or internal reflections/scattering, but seems more pervasive than that.  Re-processing the RTI without the reflective images might help provide a clue.  Does the blurriness only occur in specular enhancement, or is it also visible in the default view?

 

To reduce the blow-outs in specular enhancement, try moving the Kd slider (RGB color) all the way to the left, and decreasing the Ks slider.  Moving the N slider to the right also will reduce the blow-outs by making the reflections more concentrated (less diffuse).  If you're interested in texture, you could also try using diffuse gain instead of specular enhancement. 

Link to comment
Share on other sites

Thanks for that really good answer, am sorry to take so much time to say it.

 

I think the use of an snoot its a logical deduction, when we think small objects requires small balls, but in my case I tried with no chance. I use instad an small slave flash.

 

In other examples I made, there's no over exposing in any of the photos, but I get that all white scenes when passing to the specular enhancement.

 

I have checked the original jpegs in this tintype example with no evidence of blurring of movement from one image to other. I think I will repetar the pics in order to compare.

 

Thanks for the advice about the sliders, I realise they help in some views, in fact.

Link to comment
Share on other sites

It isn't unusual to have specular blowout when viewing RTIs in Specular Enhancement mode - particularly when the light is toward the center (high noon)  One thing to remember is that the light controller is actually moving the light in a hemisphere in the virtual viewer environment, so the center is like the sun at high noon and the edges are like the sun at the horizon.

 

You can take the middle slider (ks) and pull it to the left to reduce the specularity (shininess) of the surface.  In addition, HSH based RTIs will always be brighter in Specular mode than the PTMs.  This doesn't mean you have bad data, it's just a function of how the algorithm is calculating the data and how the viewer shows it.  You can also pull the bottom slide to the right to reduce the highlight size on the surface.

 

Related to Taylor's note about rebuilding with less images.  We do this kind of thing for tricky subjects that have a very shiny surface.  I'll note that the next release of the RTIBuilder is going to allow a new facility for managing which images are used to build an RTI, to make it way easier to do this kind of experimentation starting from a single set of images.  You will build using the maximum number of images and then you can remove images and add them back in using check boxes on thumbnails of the images from the lat screen.  All the data will be kept in a single log file, and it will be easy to reprocess without having to make copies of Jpegs, etc.  We are modifying the log file to make it easier to see exactly what data was used and with what settings, etc for each RTI built from a capture set.  You can figure a lot of this out with what is there now, but this will be much improved and easier to read, as well as having this new flexibility.  Sorry I don't have a release date at this point, though we are working to wrap it up.  One issue with the software updates is that we have had really tiny budgets to do them, so it ends up being a volunteer effort to complete them.  Add in summer travel, and things take longer.  If anyone wants to help support the updates and the general support for RTI tools, documentation, and this user forum, Cultural Heritage Imaging is a nonprofit in the US - and we totally appreciate and rely on donations. Please consider helping out.

 

And finally - it is almost impossible to detect slight movement in the image sets by looking at the images.  You can usually tell by seeing the image in the RTIViewer go blurry with certain light directions.  The team we are collaborating with at Princeton has developed an image alignment tool specifically for RTI data sets (other forms of alignment weren't really meant for this purpose)  I don't have a release date on that either - though we have been testing this software and it's very useful.  We have some loose ends to tie up, and the work gets fit in around a larger project we are doing with Princeton.  

 

Carla

Link to comment
Share on other sites

I thought it might be worth commenting a little further about how some RTIs appear "blown-out" in the RTIViewer when the virtual light position is set to high angles (>60 degrees) with specular enhancement. 

 

At the beginning of the capture sequence, the exposure is checked for both a low-angle and a high-angle light position at about 15 and 60 degrees, respectively.  It's worth keeping in mind that the RTIBuilder and RTIViewer interpolate virtual light positions for angles higher than 60 degrees.  Some RTI practitioners like to "expose to the right," meaning they prefer to be slightly over-exposed rather than under-exposed, to make use of the entire dynamic range of the camera sensor.  However, this means that the RTI will appear over-exposed for virtual light positions higher than 60 degrees.  This effect becomes more exaggerated if the surface is highly specular. 

 

I've also noticed when viewing some RTIs in specular enhancement or diffuse gain that the light distribution appears concentrated on just a part of the surface, instead of providing even illumination over the entire object.  I'm curious why this would be, because during the RTI capture sequence, the illumination is supposed to be diffuse and should cover the entire surface as evenly as possible.  Does anyone have any thoughts as to why the illumination sometimes appears concentrated and uneven in the RTIViewer?

Link to comment
Share on other sites

  • 2 weeks later...

I remade the pictures of the tintype in the video and the "problem" seems to dissapear in some level. I did that since it had become more easy than compare the 40+ original images individually.

 

I made a rti of a negative surface with the hope of transmit the importance of the light angle when confirming the non existance of fungus, for example. But even if I think I used a bad example (while the fungus where a little too visible in more than one angle) the "default/cenital" image the rti file displays seems to be a little mix with at least one or two of the others. Is that possible in the case of the first image?

 

And, of course, I consider to donate to CHI. I am not in the best economical shape right now but in the moment that changes (or if someday I use RTI in a paid work -by the moment is only for educational purposes-, whatever comes first) I will do it.

 

I can keep posting videos of RTI files (will love to have a own website based in wordpress in order to be able to put RTI files directly for web viewing) of different photographic surfaces if you want to.

Link to comment
Share on other sites

Dear MJM,

 

I think from your comments that maybe you don't understand how an RTI is calculated, and therefore you are misinterpreting the results.  AN RTi is NOT a composite image, and the viewer does not have access to the original images once the RTI is created.  I recommend taking a look at the RTI page on the CHI website, and in particular the "how does it work" section, which describes surface normals.  Essentially the RTI file contains for each pixel RGB values for color and a mathematical description of a surface normal (per pixel).  This combination of color and shape information is very powerful and allows the viewers to do the relighting and the other mathematical enhancements.  The information is calculated from the full set of images (the mathematics are available in published research papers - if you are interested) and then the original images are not used in the viewer environment.  You should archive your images though because new algorithms are being developed, and new techniques such as the Algorithmic Rendering will be able to work on these data sets, and will need access to the original images.

 

Taylor - you ask a great question.  I have also seen this phenomenon.  I don't have a great answer for it, just some observations.  I'll see if I can tap one of the geekier more technical people to weigh on this question.  My observations are that if you have a fairly rounded object, like a cylinder, a vase, etc, you will see this affect, because there aren't normals around the edges in areas that are out of focus (due to depth of field)  I've also noticed it in objects where I wasn't able to get far enough away with the light (i.e. a minimum of 2X the length of the object you are imaging) and therefore don't have good normal data around the edges, (you will have good normals in the center, but less as you move farther away from the center, due to the lack of light standof)  I'll also note that the results are a bit different depending on whether you use PTM or HSH algorithm.

 

Hope this is useful!

Carla

Link to comment
Share on other sites

I also want to add a note about donating to CHI - since I brought it up, and MJM responded.  We understand that not everyone is in a financial position to support our work. We will never have everyone who uses RTI support us, and that's OK.  However, as an independent nonprofit organization in the US, we need support in order to do this work.  We are able to get some grants and we bring in some money through training and consulting projects, but those do not cover all the costs of our organization.  We try to keep our costs as low as possible by using lots of volunteer help, engaging graduate students, and by several professionals working on projects for substantial discounts over what they charge their other clients. Still, it takes money to keep things going.  Whether or not you are engaging in commercial work, if you use RTI and you think the tools, user guides, and this forum are valuable, then we hope you will consider making a donation to keep it all going.  

 

Thank you!

 

Carla

Link to comment
Share on other sites

Carla,

 

Indeed, I don't really know too much about how RTI algoritms works. I got attracted to RTI since its based in multiple lights angles vision, thing that I always finded essential to the understanding of objects and their different deterioration factors. I am doing maybe the inverse way, starting from the pictures and facing the conceptual RTI exclusive elements later. Am sorry if I arrived to the point where I should read more instead of asking, is just that when I speaked of "the "default/cenital" image" in a negative I just made, it seemed to me that some information was lost and some where obviously "bad placed" as the effect where not the one we will see of a cenital view but from a mix between a cenital and a little rack one, and I think in this case am, again, facing a mistake I made.

 

I will keep reading and examining the original shots so.

Link to comment
Share on other sites

mjm -  Please keep asking questions!  That's what the forums are for!

 

I think there are a lot of folks who are learning and have misunderstandings or don't quite understand what is going on, so myself and others here are happy to help provide answers.  Where I can point to materials that already exist, I'll do that to save everyone time and energy.  However, that doesn't mean it's a bad question.  We certainly don't expect that everyone has read everything that is written on RTI to use it or to participate here.  That would drive a lot of people away, and our goal is to have more and more people use RTI.

 

Cheers!

 

Carla

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...