User talk:Jarmniku

Purple fringe via IR
If you have a source with that theory, let's use it. But it seems very unlikely to me, especially since the sky scatters little or no IR, while it scatters a lot of short wavelength (that's why it's blue). Dicklyon (talk) 17:58, 23 July 2020 (UTC)


 * Well, where are your sources then? Did you check my references? I do understand if you feel like I am attacking your page, sorry about that. But actually I have been thinking this for a long time, starting from photos taken with the very first Nokia camera phone, where photo of a burning fireplace appeared bright BLUE! That was very strange. Then later on, my girlfriend took some bird photos at Gran Canaria with a mobile phone, through binoculars, and that purple fringe appeared very wide around the white birds. One day, I started to think could it be infrared, and then it just seemed to make sense with all those filter and sensor curves.
 * A couple of weeks ago we bought a pretty expensive Canon EF 400 mm f/5.6 lens, and found the same phenomenon, although not that bad. However, we are a bit disappointed and I am trying to find the root cause. Probably a high quality filter can correct that, although they too cost money. Anyway, it would be nice to know the true reason for the fringing.
 * If it's ultraviolet, I am happy with that, but I would like to see better evidence about that. At least I had a real photo taken with Canon 7D, and you can easily get the same result with e.g. mobile phone camera. I actually yesterday checked with my OnePlus 7T and yes, the purple is there when shooting directly into remote's infrared LED.
 * Here is some old discussion proposing mainly infrared: https://www.dpreview.com/forums/thread/702134?page=2
 * --Jarmniku (talk) 19:25, 23 July 2020 (UTC)


 * I had just found that dpreview post, too. It appears to be the only place on the web with "IrCA".  It's too bad he didn't do a test with sky after he got the IR-cut filter.  There are quite a few dpreview threads reporting good results with a Haze 2A filter (which is the strongest of the UV-cut filters, with 50% point about 420 nm).  And it's in this book (in German; With the E-M5, which has the same sensor as the E-P5, attempts have been made to eliminate the purple fringing of Panasonic lenses with an upstream UV filter. Filters such as the Tiffen Haze 2A were used to block the entire UV component up to 405 nm. is all I can see).  I admit it's not all well sourced, but since I was chief scientist at Foveon and investigated this stuff, and was probably the one who first suggested the Haze 2A filter to our dpreview friends, I have a pretty good idea about what's real here.  Lenses typically have an axial CA that goes to hell rapidly around 400 nm, while they're pretty well behaved at near-IR wavelengths.
 * I don't think you had added any references. I didn't study all the embedded external links; let me know if some of those support your theory.
 * Check out Fig. 7b in this doc, which shows apochromat axial CA going off the rails at short wavelength. Or a more basic achromat here.  The IR-cut filters in RGB cameras necessarily cut out something like 90% at 700 nm and 99% of the near IR light, but there's generally no explicit cutoff on the shortwave end, which is why you see PF in sky and reflected sun glints and such (when only the blue sensor responds, that looks like a violet color, which gets mapped to purple in typical RGB spaces).  Dicklyon (talk) 23:22, 23 July 2020 (UTC)
 * Here is a source with info about Haze 2A: "UV Haze-2A: This filter absorbs virtually all UV light. It reduces haze more than the Haze-1. Used in high altitudes and along bodies of water, the Haze 2A is also ideal for aerial shots, mountain scenes and marine scenes. It maintains color and image clarity." This is all from before silicon sensors came into wide use, unfortunately, so they don't recognize what a great filter this is for PF.  See plots at this discussion. Dicklyon (talk) 03:05, 24 July 2020 (UTC)
 * And IR skies are dark. See Infrared photography. Dicklyon (talk) 03:47, 24 July 2020 (UTC)

Interesting... I have actually seen two clearly different types of "purple" fringing. The first one typically appears around specular reflections caused by sun rays hitting a shiny surface, and that fringing is more clearly purple. The other one is common when shooting towards bright sky, and in that case the color is much more like violet (like in that horse photo). Maybe the purple one is dominated by the IR (looks very similar to IR-LED experiment), while the violet is mostly UV? It might bring more light into the subject if analyzing these two cases separately.

I think this could be very good link into the "Purple fringing" page: https://en.wikipedia.org/wiki/Diffuse_sky_radiation It shows the spectrum that is radiated from the cloudless sky, which is actually very light blue. For me, it looks like ultraviolet and infrared are appearing about the same level (the second picture below the first one), or do I read it correctly? Funny enough, there is a photo having very strong violet fringing by change, although it is there for demonstrating other thing: https://upload.wikimedia.org/wikipedia/commons/3/3f/Canopy.jpg

There is also some commenting that cloudy sky will scatter all wavelength quite evenly - well, that's quite apparent as cloudy sky is gray.

What comes to lens aberrations, you are right, there seems to be much more aberration at UV side.

The problem with this kind of reasoning is that we all tend to see what we want to see... to support our own opinion (the confirmation bias). --Jarmniku (talk) 08:02, 24 July 2020 (UTC)