UV with the Sony DSC-F828

[Fandyus]

I have recently joined what seems to be a new wave of people buying the Sony DSC-F828 2003 point and shoot camera, mostly due to its one of a kind sensor and extremely easy full spectrum conversion.

I have been very curious what UV looks like with this camera, given its RGBE sensor. I have only obtained a fitting magnet to flip out the hot mirror filter last evening so there was no time to do UV under natural light. I got to do it now, though.

The image is in full resolution, so feel free to click and enlarge.

ZWB2+QB39 Tangsinuo stack

3s exposure, ISO 64, f/2.5

These flowers are UV yellow, yellow in real life as well. They're about as UV yellow as gerberas or dandelions. Here they are a very desaturated shade of orange. I think it's due to the fact that the Zeiss T* Vario-Sonnar 2-2.8/7.1-51 zoom lens does not pass much UV at all. That is to be expected, but still a shame, since this sensor could have unlocked a lot more UV false color, given its 4 color channels as opposed to the usual three.

Last observation: flipping out the hot mirror seems to make the image only marginally brighter, meaning the hot mirror either does not block much UV at all or the lens passes so little most of what does pass is around a wavelength the hot mirror was not designed to block. Either way, I am happy I have this camera now, the IR results have been a lot less underwhelming, and even the normal visible photos have an interesting look to them, very much unlike the images taken by modern cameras. At the risk of sounding like a huge cliche, they do look somewhat film-like color wise. Probably since back then, digital was still considered to be a replacement for slide film (and was designed as such), not a universal best way to record anything.

[Fandyus]

I have decided to also upload the RAW if anyone is interested.

DSC00295.SRF

[colinbm]

This is looking good Fandy
The T* in the lens' name could indicate the limited UV reach ?
" Carl Zeiss’s T* coating on optical surfaces ensures higher light transmission and, at the same time, reduces flare and ghosting. "

[dabateman]

Interesting that UV passes at all through that lens, and at those settings.  3S at ISO 64, does seem bad. With new denoising software,  you might be able to push that up higher, for faster shutter speeds. 

 

This also got me interested in looking at old Nikon 5700 cameras with the CMYG sensor. 

[Fandyus]

17 minutes ago, colinbm said:

This is looking good Fandy
The T* in the lens' name could indicate the limited UV reach ?
" Carl Zeiss’s T* coating on optical surfaces ensures higher light transmission and, at the same time, reduces flare and ghosting. "

Thanks, Colin.

Yes, that is what I was thinking. It is also a zoom lens so one would expect a lot of thick and/or cemented elements.

[Fandyus]

17 minutes ago, dabateman said:

Interesting that UV passes at all through that lens, and at those settings.  3S at ISO 64, does seem bad. With new denoising software,  you might be able to push that up higher, for faster shutter speeds. 

 

This also got me interested in looking at old Nikon 5700 cameras with the CMYG sensor. 

I was actually thinking that that is an okay exposure time truth be told. When I use my Industar 50-2 at f/3.5 with the same stack, mounted on my Sony a6000, I belive I get similar exposure times. Keep in mind it was fully overcast when I took this image. Broad sunlight might be better.

[Fedia]

I compared the f828 and my canon in terms of IR sensitivity. When both are used in full spectrum, adding a 720nm filter cause the sony to need slower shutter speeds to achieve the same exposure as the canon.

 

This can mean that the sensor itself is less sensitive to IR than others or/and that the lens absorbs more IR than others (it is a very powerful zoom and probably also a parfocal one so I assume it has a complicated design)

 

I was serching for a chart showing the design of the lens. I found nothing concerning the lens but this came across instead :

 

 

https://www.researchgate.net/profile/Camille-Lelong-2/publication/26547944/figure/fig3/AS:214183908429827@1428076831967/Spectral-relative-response-of-a-the-three-Canon-EOS-350D-and-b-the-four-Sony-DSC-F828.png

 

I think it's from this : https://hal.inrae.fr/hal-02590559/document

 

I have not read it yet but will do. I don't know if we can trust it since the emerald channel is called "cyan" which is clearely not the same.

 

 

[Andrea B.]

Thanks for that raw. I will enjoy loooking at it. I'm still not entirely sure how to run an SRF through Raw Digger.

 

This camera is quite interesting. The chart shows that there isn't much difference between the "emerald" and green peaks. The "emerald" is, however, closer to cyan. I'm guessing 510 nm and 520/525 nm in the chart? 

Cyan ~ 500 nm. Green ~ 550 nm. The nm for the colors seems to depend on which "authority" you read.  I just ran a quick Google for those values.

 

Given the age of the F828 sensor and not necessarily UV-capable lens, etc, I also think 3 sec isn't too long for overcast skies. 

 

I did *not* realize that the Nikon 5700 had an unusual sensor !! Gotta look that one up.

[GaryR]

Apparently, the Color filter Array (CFA) in the Sony DSC-V1 was also RGB+E.

 

DPreview specs refer to it as G-R-G-B, and this site mentions it as RGB+E.
http://www.digitalsecrets.net/Sony/AdvancedKnow3-noise.html

 

Oddly enough the next gen DSC-V3 had a standard RGB CFA.

[dabateman]

12 minutes ago, GaryR said:

Apparently, the Color filter Array (CFA) in the Sony DSC-V1 was also RGB+E.

 

DPreview specs refer to it as G-R-G-B, and this site mentions it as RGB+E.
http://www.digitalsecrets.net/Sony/AdvancedKnow3-noise.html

 

Oddly enough the next gen DSC-V3 had a standard RGB CFA.

According to Wikipedia,  the RGBE was only used with the ICX-456 8Mpixels sensor:

https://en.m.wikipedia.org/wiki/RGBE_filter

 

So it might only be in the F828 camera.

 

The CYGM color array was used in many cameras. At least 16 Nikons a bunch of Canons and Panasonic and Leica compacts:

https://en.m.wikipedia.org/wiki/CYGM_filter

 

Sadly only the Lodestar X2C was the only lens interchangeable camera with that sensor (C-mount). Its low resolution and a pain to work with. But really UV sensitive. 

[Andrea B.]

Nikon USA does not show a Nikon D5700 ??? Was this body sold in Europe or Asia ?

 

Nevermind. The 5700 was a Coolpix and I didn't go there. Found it now.

[Fedia]

Would a 5700 coolpix be easy to convert to full spectrum ?

[dabateman]

1 hour ago, Fedia said:

Would a 5700 coolpix be easy to convert to full spectrum ?

Not sure but, found a bunch of repair manuals online: 

https://www.manualslib.com/manual/735323/Nikon-Coolpix-5700.html?page=13#manual

 

[Andrea B.]

Here is the Raw Digger analysis of Fandy's UV photo.  The UV reflective areas (brighter areas) are being recorded mostly in the blue channel. Remember that a difference stack of 4 different channels is going to need an inversion in order to present "normal" colours.

 

What we do not know is whether the blue channel predominates because:

• the F828 and its lens cannot record below, say, 370 nm ?
• the Bayer dyes are different in the F828 ?
• the sensor has a different sensitivity as compared to more recent CMOS sensors ?
• Could be some combination of the 3 possible factors.

 

4-channel raw composite from raw digger

 

inverted 4-channel raw composite

 

 

red channel

 

green channel

 

blue channel:  This is the brightest channel.

See histogram below for confirmation of that.

 

emerald channel

They had to name that channel something. And I'm guessing that either "slightly blue-ish green" or "cyan-ish green" would have been a bit difficult to use in marketing. "Emerald" sounds classy.

 

 

Raw Histogram:  Note blue channel.

The UV-reflective parts peak around EV1.

 

jpg extract from the raw SRF

[Fandyus]

@Andrea B. Thanks for the analysis. I appreciated seeing the data.