Color Filters to Improve IQ in Poorly Corrected Optics

A forum to ask questions, post setups, and generally discuss anything having to do with photomacrography and photomicroscopy.

Moderators: rjlittlefield, ChrisR, Chris S., Pau

apt403
Posts: 37
Joined: Mon May 06, 2019 5:29 pm
Location: Yelm, WA

Color Filters to Improve IQ in Poorly Corrected Optics

Post by apt403 »

My line of thinking is as follows: One of the big problems with lower quality objectives and lenses is the poor correction for CAs. If one were to buy a set of red, green, blue color filters and take 3 monochromatic images (after stacking) w/ each filter, then Photoshop could handle aligning them. Set each image to the appropriate RGB color channel, and potentially one would have an image that's more well corrected than would be possible with imaging in one shot. It's certainly a popular approach in the astrophotography world, where lots of telescopes are fitted with filter carousels over B&W CCD sensors that excel at low light capture.

Going a little farther back, Sergey Prokudin-Gorsky did some amazing color photography work back in the early 1900s w/ a similar technique.

Has anyone tried something similar? I'm researching filters w/ appropriate passbands now that don't cost an arm and my first born (unsurprisingly frequency response is poorly reported with cheaper filters). I'll post the results here once I've picked the filters up and given it a shot, but I figure if anywhere, someone on here has had a similar thought!

zzffnn
Posts: 1896
Joined: Thu May 22, 2014 1:25 pm
Location: Houston, Texas, USA
Contact:

Post by zzffnn »

Green filter is routinely used for achromatic microscope objectives. Many cheap green filters are good enough.

Many microscopists photograph live fast- moving subjects, so combining multiple filtered images is not very practical.

Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

That's an interesting thought, apt403. Astrophotographers do this regularly, but with monochrome cameras. I have seen some shots of terrestrial objects using this technique, and sometimes they look natural. Other times, not so much.

When I saw the subject heading, I imagined you were going to suggest filters that cut out the extremes of the spectrum and left the middle wavelengths. Many lenses have red or violet fringes. Astrophotographers and some macro photographers (I think I recall JH posting about this) have used filters that cut out the far red and the violet. THis can remove the color fringes. There is a whole industry of "fringe-killer" filters made for astrophotography. I am going to buy a set and play with it. I currently often use an IR-cut filter for macro work and I think it helps, though I don't have any test results handy.

Anyway, look up "astrophotography minus violet" and "baader fringe killer" to enter the very diverse world of filters to remove CA.

apt403
Posts: 37
Joined: Mon May 06, 2019 5:29 pm
Location: Yelm, WA

Post by apt403 »

zzffnn wrote:Green filter is routinely used for achromatic microscope objectives. Many cheap green filters are good enough.

Many microscopists photograph live fast- moving subjects, so combining multiple filtered images is not very practical.
Good point! I'd like to give that a try, as well!

Very true, this would only be applicable to prepared, immobile specimens, unfortunately. Perhaps one day someone will figure out something like a liquid lens combined with enough computing horsepower to do stacking in real time, output to a monitor. It's kinda nice we're at a point as a species that a lot of our challenges are primarily limited by budgets and engineering, and not by being confounded by physical principles that have yet to be elucidated.
Lou Jost wrote:That's an interesting thought, apt403. Astrophotographers do this regularly, but with monochrome cameras. I have seen some shots of terrestrial objects using this technique, and sometimes they look natural. Other times, not so much.

When I saw the subject heading, I imagined you were going to suggest filters that cut out the extremes of the spectrum and left the middle wavelengths. Many lenses have red or violet fringes. Astrophotographers and some macro photographers (I think I recall JH posting about this) have used filters that cut out the far red and the violet. THis can remove the color fringes. There is a whole industry of "fringe-killer" filters made for astrophotography. I am going to buy a set and play with it. I currently often use an IR-cut filter for macro work and I think it helps, though I don't have any test results handy.

Anyway, look up "astrophotography minus violet" and "baader fringe killer" to enter the very diverse world of filters to remove CA.
Thanks! You've got my curiosity piqued! My main 1:1 lens, a Tokina 100mm f/2.8, has some unfortunate issues with LoCA, those filters could be quite handy, especially when I've got the 2x teleconverter attached.

Since we're on the subject of astrophotography - Have you ever heard of anyone trying to adapt a field flattener to an objective or macro lens? They seem to be almost universally used by the serious astro folks, who may hate vignetting and edge distortion with at least as much, if not more, passion than the photomacrography crowd.

Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Have you ever heard of anyone trying to adapt a field flattener to an objective or macro lens?
I've been curious about that myself. But it would only be useful when we are using "thin" lenses as tube lenses, like many of us do with Raynox and other close-up filters. I think these could benefit greatly from a field flattener. There are some generalized field flatteners; I saw one with six elements which seemed very good. But they are meant to accept the output of a distant "thin" doublet or triplet lens; a complex modern camera lens like a Tokina 100mm macro already has a field flattener built into it.

I've been thinking about telescopes as tube lenses for a while, and if I ever go that route, I would make sure to get a field flattener. But most astophotography field flatteners seem too simple compared to modern camera lenses. They often consist of just two elements (the six-element one I mentioned above was very hard to find). That's because astrophotographers put a high priority on light-gathering ability and signal-to-noise ratio, since their signal is so weak. Every air-glass surface loses some signal, so they weigh the costs vs benefits of additional elements differently than we would in macrophotography. Our camera lenses probably do a better job of flattening than typical field flatteners used in astophotography.

While I was researching telescopes versus telephoto lenses, I found tests comparing a high-end telescope and a high-end Canon supertelephoto on terrestrial targets. The telescope won by a wide margin in the center, but the telephoto won by a wide margin on the edges:

http://www.samirkharusi.net/televue_canon.html
and
http://www.samirkharusi.net/tv60_canon.html

It is not clear what kind of flattener was used in the first link (mention is made of a field reducer) but in the second link a field flattener was used.

The filters I mentioned won't help LoCA at all, just lateral red or violet or purple CA. Your solution would work better. Another approach is to use three monochromatic R G B light sources like some scanners use. The ultimate solution is to go to black-and-white photography, using a single wavelength laser and a cooled monochrome astrophotography camera. That works very well if you can accept monochrome pictures.

ray_parkhurst
Posts: 3412
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

apt403 wrote: Thanks! You've got my curiosity piqued! My main 1:1 lens, a Tokina 100mm f/2.8, has some unfortunate issues with LoCA, those filters could be quite handy, especially when I've got the 2x teleconverter attached.
For LoCA, the problem is not one of alignment, but of focus. The colors are not all focused in the same plane. So I'd presume the process would be to shoot with R filter, refocus with G and shoot, then refocus with B and shoot, and then somehow combine these into a final image? I'd think you could do a similar thing without using the filters, just doing 3 shots at the 3 focal planes and then combine the appropriate color channel from each shot into a final composite.

For LaCA, the problem is also not one purely of alignment, since the center of the image often has no LaCA, yet it gets wors as you go farther from center. Available software is pretty good at correcting LaCA. I'd bet you could do a LoCA correction from 3 focal planes, then combine the 3 images and do a LaCA correction, all without having to use filters.

Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

Ray, there is a lot of cross-talk between colors in a bayer sensor's R, G, and B color channels, and I think this would make problems for methods that depended on using white light and then separating color channels.

apt403
Posts: 37
Joined: Mon May 06, 2019 5:29 pm
Location: Yelm, WA

Post by apt403 »

Sorry for the lack of quotes, didn't want to make a hugely lengthy post:

@ Lou: Ahh, that makes sense. Yes, I suppose if most field flatteners are only two elements, optical performance would suffer unacceptable losses. That six element unit seems interesting!

Thanks for the links! Interesting stuff.

Haha, after posting my last reply, I was doing some more research and came across your monochrome work with the Ultra Micro Nikkors coupled with the cooled CCD astro sensor and my jaw dropped. That's some truly amazing stuff! I've never seen 400% crops with such detail! :shock: Thanks for that thread - I've now got some idea of where I need to head in the future. :D

@ ray_parkhurst: I see, that does change the nature of the problem, and I believe I'm experiencing exactly what you describe. I ran some very quick tests last night, comparing an image captured using a composite white light source (RGB LEDs), versus three images illuminated with only the red, green, and blue light. The three images were then stacked and aligned in PS.

I need to rerun this test after dark tonight when I can again block out all the ambient light sources, but my initial results are interesting. Color rendition seems to be much improved on the composite image, along with lesser gains in apparent contrast and resolution. What I wasn't expecting (prior to reading your reply) is that the in-focus area of the composite image varies substantially from that of the single shot, white light image. This very well may be experimental error. I'll post my revised results here tonight.

ray_parkhurst
Posts: 3412
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Lou Jost wrote:Ray, there is a lot of cross-talk between colors in a bayer sensor's R, G, and B color channels, and I think this would make problems for methods that depended on using white light and then separating color channels.
I suppose that's true, plus the Bayer filter array has very poor color resolution.

Edited to add: but putting a filter over a Bayer array is also problematic. I see the process working for monochromatic sensors, but if the array already has a filter array, how much improvement can you achieve? Will be interesting to see any results.

cube-tube
Posts: 107
Joined: Tue Oct 10, 2017 7:50 am
Location: Durham, NC

Post by cube-tube »

Maybe I am missing something--wouldn't RGB channels photographed separately be just as misaligned as they would be from a combined (white) light source?

I wouldn't think that separating the channels would have any effect on the achromaticity of the lens.

I can see how a single monochrome image would be an improvement over a color image, though.

Smokedaddy
Posts: 1951
Joined: Sat Oct 07, 2006 10:16 am
Location: Bigfork, Montana
Contact:

Post by Smokedaddy »


Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

M
aybe I am missing something--wouldn't RGB channels photographed separately be just as misaligned as they would be from a combined (white) light source?
If you rescaled two of the three channels to align with the third, then they would remove lateral CA.

ray_parkhurst
Posts: 3412
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Post by ray_parkhurst »

Lou Jost wrote:M
aybe I am missing something--wouldn't RGB channels photographed separately be just as misaligned as they would be from a combined (white) light source?
If you rescaled two of the three channels to align with the third, then they would remove lateral CA.
There are already pretty good algos to do this with a single image, so what is the advantage to adding filters?

Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

If you do this with a single image, you are accepting crosstalk between the channels. In other words, some of the energy in green channel is actually produced by blue light (that is not necessarily focused in the same plane as the green light).

Lou Jost
Posts: 5943
Joined: Fri Sep 04, 2015 7:03 am
Location: Ecuador
Contact:

Post by Lou Jost »

If you do this with a single image, you are accepting crosstalk between the channels. In other words, some of the energy in green channel is actually produced by blue light (that is not necessarily focused in the same plane as the green light).

Post Reply Previous topicNext topic