I Gilded the Lily

[colinbm]

Yes I can send the 3 DNG files to you, can you PM your email so I can send them by Drop Box, please.

[dabateman]

Very nice. An excellent example that deeper uv is different. I have seen some differences, but its always hard to remember. Now we have an example to show Andrea when she says she hasn't seen one.

 

 

[colinbm]

Thanks Dave, I'll get some that are better. All the in camera WB were pretty close. My PTFE has a very rough surface from when I cut it on the bandsaw, that is making it easier to expose the image as I can see when the rough surface blows out.

[Stefano]

So, here's a TriColour stack with Colin's first series. I desaturated the images in IrfanView, and made them red, green and blue in IrfanView too (in the Color correction window). Red = 395 nm, green = 365 nm, blue = 340 nm. I then stacked the images with Image Stacker. In the final result, the PTFE block was white, but I still preferred to do a click white balance in IrfanView.

 

The result:

 

Green anthers. Green = 365 nm, so yes, they have a bump there. Very interesting.

 

There is some uneveness in the lighting, but it isn't very important (unless you want an accurate image).

[colinbm]

Thanks Stefano, this is really funky....

I'll try & get the lighting more even.

[colinbm]

I have redone the three pictures, still with the same three LEDs.

I changed the Lens to the O D Industries 40mm enlarger lens, it probably transmits slightly lower into the UVA. I also placed the BaaderU filter behind the lens inside the tube.

Seems to be all very different....

 

340nm LED light.

 

 

 

365nm LED light.

 

 

 

395nm LED light.

 

[Andy Perrin]

Here's what I get from the second set. This one has no white balance, it's just the 3 images aligned and put in the RGB channels.

 

Without the PTFE in the shot I didn't know how to white balance it properly so I tried clicking aorund in a few spots. Colin, the LEDs seem to have moved a lot between shots because some areas have lots of one color vs other areas have lots of a different color.

[Stefano]

Colin sent me his .DNG files, and I did some experiments. I will post the best ones (the ones done with the best techniques). This is the second set.

 

For TriColour, I first processed the images in Photo Ninja by removing any tick (color, exposure, sharpness etc.), and converting them in B&W. After that, I converted them into "color channels" in IrfanView. In the 395 nm image, the correctly exposed portion of the PTFE block is not in the same spot as the portions in the other two images, so I couldn't simply stack the images and do a white balance. Instead, I checked the colors in Paint (a random pixel in the correctly exposed regions, I didn't do an average), and readjusted the brightness of each image in Windows photo editor so that it was about the same. I checked again with random pixels in Paint and the variations are minimal. I then stacked the images with Image Stacker, without any further edits. Colors as before.

 

Here it is:

 

Then, since I had the raw images, I tried something I did try myself in IR once. I removed any color corrections, so that the raw colors were displayed. I then saved the images as .jpg, stacked them and white balanced the resulting image in Photo Ninja. This simulates a single exposure with all three LEDs, with about the same sensitivity for the three wavelengths (in real life the 340 nm illumination would be almost unnoticeable). The result:

 

Once can notice the wider color palette, as well as the typical blue-yellow palette in the PTFE block.

 

Perhaps there are better ways to do what I did, this is what I could do.

[Stefano]

Andy, I can send you the raws if you want them.

[Andy Perrin]

Interestingly, if you just take that 340nm image and boost the saturation/vibrance and contrast a bit, you get a nicer looking "tricolor" than the actual tricolor.

[Stefano]

Agree, that's really pretty.

 

In my "raw stack" image (the second one) you see the lighting unevenesses much less.

[Andy Perrin]

Agree, that's really pretty.

 

In my "raw stack" image (the second one) you see the lighting unevenesses much less.

Yeah, if you have the raws, that was definitely the way to go. The only quibble I had with your process was this bit:

 

"I then saved the images as .jpg, stacked them and white balance."

 

You shouldn't have converted to JPG prior to white balancing. The JPG degrades the colors and the resolution. It's best to save conversion to JPG for the very last step, and work with TIFFs or PNGs, which are uncompressed (or lossless compression).

[Stefano]

I don't know how to stack raw images. The software I have may handle this, but I never tried.

[Andy Perrin]

You don’t stack the RAWs, you save as TIFF and stack TIFFs.

[Stefano]

Thanks Andy, now I understand how to do this. I just checked and Photo Ninja can do this. Very useful.

[Stefano]

It seems my software can't stack TIFFs. I think it can accept .tif files.

[Andy Perrin]

Those are the same. It’s merely a different extension on the file name. In the old days, Windows would only allow 3 letters.

[Stefano]

Perfect. Thanks again. I'm going to te-try my second experiment that way.

[Stefano]

I have a problem I have to understand. The image comes out corrupted. If I can solve it I will post the result.

[Stefano]

The software gives me errors like this: "Warning, unknown field with tag 34665 (0x8769) ignored."

 

The resulting image has black vertical lines and other errors. I don't know if this is a common issue, I never worked with .tif. I don't know if someone can help, it can be a software-specific isuue.

[Andy Perrin]

I have never seen that. I think it’s the software?

[Stefano]

It's odd, since apparently it is designed to handle .jpgs and .tifs. I don't think Photo Ninja is the problem, unless there's some settings I didn't use. Very odd.

[Andrea B.]

Note about stacks...

After making the stack, it is sometimes useful to use Auto-Levels (works to balance each channel separately) to improve the result. If it looks good, keep it. If Auto-Levels looks bad, undo it.

Another trick is not to use Auto-Levels, but to balance the end points of the brightness histogram. This is alternately known as setting the black point and the white point.

 

---

I have occasionally had a problem where a TIF file saved in one program is refused by another progrem. The fix is to re-save the TIF file in some third program whose output is readable by the program which rejected the original TIF.

 

Example: If I export a TIF from Raw Digger, then Photo Ninja will not accept it. So I resave the Raw Digger export in Photo Mechanic (my viewer/sorter). Then Photo Ninja will accept the resaved TIF.

 

I could probably sort that out. But the re-save trick worked so I never have.

[Andy Perrin]

I tried that, Andrea (in my attempts anyhow). I found the effect isn't much different from a click white balance on various plausible-looking colors. I think Stefano was on the right track with using the original RAW with no white balance at all on each of the three wavelengths. The last image he posted using that method looked most like what I think the result should look like, although it needs to be white balanced still using un-JPEGed colors.

[Andrea B.]

yep, sometimes it is useful, sometimes not.