rfcurry Posted September 30, 2022 Share Posted September 30, 2022 I have been thrown a bit by some false colors I am getting with some filters. The three photos below were all taken with a Panasonic GF1 full-spectrum-modified, a Schneider-Kreuznach Radiogon 4/35 (4 elements/4 groups) lens, f8, ISO 800, the Vis exposure was 1/800s, the two UV bandpass filters were 2s. Overcast day, mid-afternoon. In-camera WB on a PTFE disc, each filter was WB separately. One-click PP WB. I was so close that I used the Vis focusing on the UV shots...laziness. :) 1st image - Vis with NIR-Block filter 2cd image is with a UV 379BP43 filter. The colors seem pretty "UV standard". 3rd image is with a UV 330BP93 bandpass filter. The colors are totally unexpected. (I also took shots with a Hoya 72 NIR pass filter over the UV filters. They showed no leakage in the NIR.) Has anyone had similar Colorchecker results in the UV? Thanks. Link to comment
Andy Perrin Posted October 1, 2022 Share Posted October 1, 2022 I have not, but the Joker here may be the camera itself. I vaguely recall Andrea saying that one of the Panasonic cameras had unusual false colors. I do not have a reference for that statement, but maybe she or another Panasonic user can confirm. Link to comment
Stefano Posted October 1, 2022 Share Posted October 1, 2022 The flowers look dark in the 2nd image but reflective in the 3rd one. If there isn't an IR leak, I would look for a visible light leak. Link to comment
Stefano Posted October 1, 2022 Share Posted October 1, 2022 Try to see if sunscreen looks like this: https://www.ultravioletphotography.com/content/index.php?/topic/3478-how-to-wb-this-or-maybe-the-filter-i-have-is-bad/ Link to comment
rfcurry Posted October 1, 2022 Author Share Posted October 1, 2022 Stephano, I went back outside after the rain paused today. To test for visible light leaks, I put the 330BP93 over a UVIR cut filter (Andover 505FD64). Here is the 330BP93 alone: and the 330BP93 over cut filter: The only PP was reducing to 1000px width. Here is the transmission curve of the 505FD64: Obviously, I need a better UVIR cut filter. From the above I see no visible light leak. Thanks for the suggestion. P.S. - all parameters were the same as the previous day. Here is the curve of the 330BP93: Link to comment
Stefano Posted October 1, 2022 Share Posted October 1, 2022 The 330BP93 + UV/IR cut photo looks UV-ish, which makes sense since it looks like your Andover filter leaks some UV in the graph. I can't say for sure where the leak is, try to take photos such as spectrums with a diffraction grating or other subjects (sunscreen, polycarbonate goggles, filters etc.), that could help. Link to comment
rfcurry Posted October 2, 2022 Author Share Posted October 2, 2022 Stefano, Having tested for leaks in visible light and NIR light, and finding none, I conclude that the colors are from the UV. However, what wavelengths of the UV produce a colorchecker like the above? I took some more shots with a different lens with a similar result. Link to comment
Stefano Posted October 2, 2022 Share Posted October 2, 2022 Look at the bottom three squares of the leftmost column (yellow, orange, red). They look very different in your UV photos. Also, the stems of the flowers are not dark in the 330BP93 photos (you can see this better in your second series). Bright surfaces that are supposed to be UV-dark strongly indicate a leak. I don't know where the leak is, but I'm sure it is there. Do you have some U-glass (U-360, U-340, UG1, ZWB2...)? If so, try to stack it on your filter. It will block all visible light and strongly reduce IR. I think it should improve your images. Link to comment
dabateman Posted October 2, 2022 Share Posted October 2, 2022 I haven't imaged my color checker in each wavelength yet, but I have seen a golden image at 335nm, depending on the white balance. I wouldn't be surprised, if Panasonic is using some none typically color dyes on your sensor, like what Andrea saw with her S1R. The typically color dyes show a distinct green in UVB. You might have different sensitivity of the red dye to make itore yellowish. Link to comment
rfcurry Posted October 3, 2022 Author Share Posted October 3, 2022 Dave, Thank you for your input! I have never used a UV bandpass filter that peaked at such a low wavelength - 330nm. Perhaps the "golden image" at 335nm is what I'm getting. When I set the in-camera WB with a gray PTFE, it gave me a mellower yellow. All parameters are as previously recorded, except the lens used was a Ludwig Meritar 50/2.9. Here is the Vis image with gray PTFE WB: Here is the 330BP93 image with a gray PTFE WB, after a single-click PP WB: And here is the 330BP93 image with a white PTFE WB, after a PP WB: The yellow cast is muted in the gray PTFE WB image. Is it similar to your results at 335nm? Thanks. Link to comment
ulf Posted October 3, 2022 Share Posted October 3, 2022 The curve of the 330BP93 reminds me of my measurements of the ZWB2 8mm and 2 x U-340, 4mm. https://www.ultravioletphotography.com/content/index.php?/topic/4785-what-is-everyone-doing-with-their-8mm-zwb1/page/3/#elControls_49506_menu The curve of the 330BP93 has similar transmission and IR bump at 700nm as the ZWB2, 8mm. I have not yet had time to try those two filters on flowers that can give yellow false colours. However, with both of them I noticed that it was impossible to make a meaningful WB, with my camera in combination. I saw that with any lens that had a limited UV-reach, even with the UV-wise the very good Focotar-2 50mm https://www.ultravioletphotography.com/content/index.php?/topic/4367-leitz-50mm-f45-focotar-2/. I think the Focotar-2 has a better UV reach than most accidental lenses we use. I got similar golden colours as you when trying to do a VB. https://www.ultravioletphotography.com/content/index.php?/topic/5562-zwb1-8mm-and-lens-uv-reach/#comment-58938 It was not until I tried lenses with an UV-reach approaching 200nm or better that WB worked the way we are used to with more normal UV-pass filters. I think that what you see here is the result of the combination of the limited sensor sensibility-slope at UV-B in combination with the UV-cutoff of the lens. That would result in a rather limited colour range in the images, making WB quite unpredictable, but mainly showing the typical false yellow from around 350nm, as light from deeper UV is cut off by sensor and lens. Link to comment
rfcurry Posted October 3, 2022 Author Share Posted October 3, 2022 Ulf, Thank you, that was quite informative. I missed the ZWB1-8mm discussion, somehow. I think you are right as to the cutoff from both lens and sensor. I don't know the UV-reach of the Radiogon or the Meritar lenses. The sample chamber of my new-to-me spectrophotometer is large enough to hold one of my 50mm lenses. I understand that an integrating sphere would be the preferred method of transmission testing, but cannot a lens, placed in the direct lightpath between source and sensor, be adequately tested for transmission? thanks Link to comment
ulf Posted October 4, 2022 Share Posted October 4, 2022 8 hours ago, rfcurry said: Ulf, I understand that an integrating sphere would be the preferred method of transmission testing, but cannot a lens, placed in the direct lightpath between source and sensor, be adequately tested for transmission? thanks By placing a lens in the direct collimated light path, the exit beam is changed so much that the measurement information gets distorted. Anything that changes the collimated beam more than a flat object like a filter, window or kyvette will cause this measurement deviation. Only by collecting all exiting beams and averaging them in an integrated sphere the result will be fully OK. You might get some indicative result of the cutoff, but the amplitudes will be way off, likely too low. Unfortunately the deviation also differs over the wavelength range, due to several optical phenomenons. Without in detail knowing the optical design of your spectrophotometer, it is impossible to predict how incorrect the results will be. Do NOT expect them to be very valid. I used a collimated beam setup in the beginning, before I got my integrating spheres and a light source powerful enough as needed, for lens measurements with integrating spheres. If you search and look at my early topics you can see that. Then I usually normalised the results to 100% at around 400-430nm to get a more comparable result. With my collimated setup it was super critical to find the optical axis of the lens and have the lens perfectly aligned along the axis. X, Y, tilt and yaw had to be quite right. To get a really good UV reach without buying an UV-Nikkor you could cobble together a primitive lens based on a fused silica planoconvex element. The image quality is not fantastic, but stopped down to typical macro aperture of f/11 or so it will be usable. The aperture could even be fixed. Link to comment
JMC Posted October 4, 2022 Share Posted October 4, 2022 Reed, I measured a couple of Meritar 50mm lenses for transmission in the early days of setting up my method. See one here - https://www.ultravioletphotography.com/content/index.php?/topic/3406-lens-transmission-in-the-uv-latest-update/&do=findComment&comment=28914 - and a second one in one of the graphs on page 2. Essentially by 320nm I would expect it to be opaque. Couple that with the likely response of the camera, and the spectrum of the light, and the UV component of the image will be mainly from the upper part of the UV range. As Ulf mentioned a simple UV fused silica element would be a good thing to try here. Link to comment
Andy Perrin Posted October 4, 2022 Share Posted October 4, 2022 Ulf/Jonathan - would a pinhole also work? Or would the exposure time be too long with such weak light? Link to comment
ulf Posted October 4, 2022 Share Posted October 4, 2022 The bigger diameter of pin hole the fuzzier picture, but also shorter exposure time. If the camera is OK with long exposure times I think it will be OK, The transmission of a pinhole or field plate is naturally even better than a FS-lens. It is worth a test to see if there are enough details left in the image. Link to comment
rfcurry Posted October 5, 2022 Author Share Posted October 5, 2022 Jonathan, Thanks for the Meritar info. I don't know how accurate Klaus' old UV range for lenses chart is, but the best of the list is a Cassar-S. I have one of those, so I'll try that next. Ulf, I'll pick up a single FS element, they aren't expensive. I can pull an iris from an old lens and jerry-rig something, I'm sure. Thanks for the suggestions. Link to comment
ulf Posted October 6, 2022 Share Posted October 6, 2022 11 hours ago, rfcurry said: Ulf, I'll pick up a single FS element, they aren't expensive. I can pull an iris from an old lens and jerry-rig something, I'm sure. Thanks for the suggestions. For the most simple solution you could just have a fixed aperture suitable to the FL of the lens element. I would recommend a FL , not less than 70mm. The range 70-110mm is most practical and high speed is not very usable a rather small lens element will work fine. Minimum configuration: Hood -- filter -- lens -- fixed aperture close behind the lens -- Extension tubes -- camera mount adapter. Focus by moving the camera. Next level is to add a helicoid. A salvaged short FL (35mm?) lens without any lens elements might be a way to get a reasonably good iris. That must then sit on a helicoid... Link to comment
Andrea B. Posted February 25, 2023 Share Posted February 25, 2023 Looks like green leak to me. The pigment in that flower is a cyanidin type anthocyanin indicating UV absorbance thru the range 300-00 nm with very little variation, if any. Thus the flower cannot be very UV dark around 380 nm and then suddenly very UV reflecting at just 50 nm away. Link to comment
rfcurry Posted March 20, 2023 Author Share Posted March 20, 2023 Andrea, I have not pursued development of the 330BP93 since I couldn't account for the UV coloration. You are right it did look like a green leak. Anyway, it was too questionable a filter design. I used some of the same dichroic glass and different ionic glass to produce the 376BP41 which has this curve and produces no weird UV colors. :) Thanks. Link to comment
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now