Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Is R3D 4:4:4 or 4:2:2?

Fujifilm EXR CMOS pattern.

They have some serious problem with white clipping in the cameras that use this, especially X10.

fujifilmexrcmosimgmix.jpg



Fujifilm X-Trans CMOS pattern. For the soon arriving X-Pro 1 camera.

fujifilmxtranscmos03sma.jpg






xtranscmospdfcapt01smal.jpg


X-Trans CMOS Sensor

The images from this sensor are very promising.

Fujifilm is also working on a CMOS with a organic layer that only let in visible light, which means they can also get rid of the IR-filter. Combining the organic layer, no need for micro lenses and filters and the new "random pattern" can become a very interesting sensor.

Fujifilm patents hybrid organic/CMOS sensor
 
Just imagine where digital imaging technology will be in 10 years time?
 
Here is a really ignorant question: since silicon chips are not as good at collecting blue light energy, why not design a pattern that samples more blue sites than either green or red?
 
Iannis Holwech said:
Fujifilm EXR CMOS pattern.

They have some serious problem with white clipping in the cameras that use this, especially X10.

fujifilmexrcmosimgmix.jpg



Fujifilm X-Trans CMOS pattern. For the soon arriving X-Pro 1 camera.

fujifilmxtranscmos03sma.jpg






xtranscmospdfcapt01smal.jpg


X-Trans CMOS Sensor

The images from this sensor are very promising.

Fujifilm is also working on a CMOS with a organic layer that only let in visible light, which means they can also get rid of the IR-filter. Combining the organic layer, no need for micro lenses and filters and the new "random pattern" can become a very interesting sensor.

Fujifilm patents hybrid organic/CMOS sensor
Click to expand...



Wow

E
 
Scott Crawley said:
Here is a really ignorant question: since silicon chips are not as good at collecting blue light energy, why not design a pattern that samples more blue sites than either green or red?
Click to expand...

Because human eyes are most sensitive to green, and so green resolution is perceptually more important for overall luminance resolution.

That Fujifilm sensor looks very interesting!
 
It does, but "prevents moire and false color" and "moire and false colors are eliminated" are very strong sentences, especially without proof in terms of at least a high resolution zone plate image to back up the "theory". I laugh that they think that a repeating pattern is somehow random.... Seriously though - I want to get one of these Fuji cameras on a chart and see what exactly is going on because quite frankly I don't think you can avoid sampling theory so easily.

Graeme
 
Graeme i dont trust you ....you sound like a man with all the answers who is having fun with old questions......... oh to be a fly on that red wall :)
 
I don't trust me either :-) Hence I measure and re-measure. It's the only way to be sure. As for fly on the red wall - you'd be mostly bored, with small periods of being bamboozled and terrified.

Graeme
 
Graeme Nattress said:
It does, but "prevents moire and false color" and "moire and false colors are eliminated" are very strong sentences, especially without proof in terms of at least a high resolution zone plate image to back up the "theory". I laugh that they think that a repeating pattern is somehow random.... Seriously though - I want to get one of these Fuji cameras on a chart and see what exactly is going on because quite frankly I don't think you can avoid sampling theory so easily.

Graeme
Click to expand...


He, he…

Graeme your "ignorance" is spell binding...

It’s powered by an imaginanium substrate… that transcends sampling theory.... (gohhh!), some people!

Fuji seems to be fond of these; but power to them to be creative in the market place. If they built a MF camera that might be interesting…

Rather telling as to what happens next (i.e first and second stage processing) if no OLPF is required…

I guess the combination of chopped up pattern + heavy smoothing and processing = nice looking lower res images.

As it is a repeating rectilinear pattern you would think there would still be pronounced diffraction effects.

A truly random array could be quite “silky” but would require massive redundancy.

I could think of a few test patterns where that sensor would have a total egg, but I guess the image is so processed that the actual characteristic of the sensor is never seen, which I guess is the whole point.

For some reason this makes me think of old movies, fading starlets and Vaseline on the lens.

When it comes down to it, it’s hard to find things that are genuinely new. At least somebody is trying...


E
 
Graeme Nattress said:
I don't trust me either :-) Hence I measure and re-measure. It's the only way to be sure. As for fly on the red wall - you'd be mostly bored, with small periods of being bamboozled and terrified.

Graeme
Click to expand...

That’s so funny…

That sounds like my week!

When will it ever end…????

Best,

Eric
 
Eric & Graeme

As a farmer / want to be reduser i can tell you there is nothing boring about what you guys do tho at times it does hurt my brain to hear you talk.....im going to bed cheers

Ev
 
Graeme Nattress said:
It does, but "prevents moire and false color" and "moire and false colors are eliminated" are very strong sentences, especially without proof in terms of at least a high resolution zone plate image to back up the "theory". I laugh that they think that a repeating pattern is somehow random.... Seriously though - I want to get one of these Fuji cameras on a chart and see what exactly is going on because quite frankly I don't think you can avoid sampling theory so easily.

Graeme
Click to expand...

I think if you grab your average British physics teacher, that is normally given to wearing a herring bone tweed jacket when shot at the right distance with the Fuji new arrangement should render said teacher in a rather psychedelic fashion… with the appearance of a grey jacket covered in Christmas fairy lights.

(elbow pads not included)…

[back to work Eric… ptishhh! (crack of the whip)]

Black and white Herring bone pattern for “best” results
 
Evan Starkey said:
Eric & Graeme

As a farmer / want to be reduser i can tell you there is nothing boring about what you guys do tho at times it does hurt my brain to hear you talk.....im going to bed cheers

Ev
Click to expand...

Well fingers crossed for a bumper crop this year for you! I have to drag my arse to the barn.. and start my day (on this hemisphere) coffee is not working...

Sweet dreams Evan!

E
 
Graeme Nattress said:
It does, but "prevents moire and false color" and "moire and false colors are eliminated" are very strong sentences, especially without proof in terms of at least a high resolution zone plate image to back up the "theory". I laugh that they think that a repeating pattern is somehow random.... Seriously though - I want to get one of these Fuji cameras on a chart and see what exactly is going on because quite frankly I don't think you can avoid sampling theory so easily.

Graeme
Click to expand...

I really hope you get to do put this against a zone chart and show us the results. :drool: I wish there where more zone chart shots published from all the DSLR stills cameras too.

The Fuji X-Pro1 APS-C, 16Mpx, mirrorless camera with interchangeable lenses, with the X-Trans "random pattern" CMOS is not in the shops yet, but there are quite a number of them "in the wild". Fuji has not released a RAW converter for the files yet, the images are only Jpeg's from the camera, so not possible to judge the quality yet.

What is interesting to note is that the "point 'n shoot" cameras Fuji released last year had the EXR CMOS sensor with the 45° rotated pattern, in 1/2" 16Mpx, 2/3" 12Mpx and APS-C 12,3Mpx cameras, and they have some serious problem in certain highlight situations (the X10 has gotten most notice) which are now called the "Orb problem".


A new firmware to fix the problem seems to fix nothing.
Why did they leave this sensor design behind?

If this 45° rotated array is something Fuji have shared with Sony, and the "orb problem" is related to this kind of sensor then Sony might get problems with the F65 that use this 45° rotated sensor pattern. Just a speculation of course, but something to look for in F65 tests.

Back to the X-Trans; Some guy tried to make a "side by side" simulation with and without AA-filter. Don't know the value of this test, but can't hurt to post it; http://sigwinch.uni.cx/post/15613562377/fujifilm-xtrans-moire-simulation

Fuji was awarded a patent late last year for their "Organic" CMOS. Here is the short White paper explanation from 2010; (pdf) http://www.fujifilm.com/about/research/report/055/pdf/index/ff_rd055_004_en.pdf
 
Iannis, thanks for the link to the X-Trans experiment. I think that experiment is good as far as it goes, but obviously we need to see it with proper demosaics on both X-Trans and Bayer so that the results can be viewed at full resolution. Even though, it's not showing elimination - only a reduction, and although it's a pattern that shows aliasing and moire, it's perhaps not the best pattern to test with. As I said above, we've got to measure to be sure!

Graeme
 
Iannis Holwech said:
I trust you to "wring the truth out of" the X-Trans sensor at some point Graeme! :wink5:

If it holds up somewhat to the Fuji claims, it is interesting, maybe even exciting. If it does not, we move on.

By the way; forgot to post this comparison between the X-Pro1 (X-Trans) and 5DII. http://dslr-check.at.webry.info/201202/article_14.html
Click to expand...

the high ISO images from the x-pro1 look interesting in that comparison, certainly less amplified moire.. i'm gunna have a play with that filter pattern design and see how it compares to a standad debayer..
 
Graeme,

I would like to know if I am OK posting some example frames or footage links here on Reduser using some advanced techniques that should provide higher detail in the footage from R3D.
I am not intending to provide or sell a tool to do this to Red camera customers. Just research at this point.

Do you see any problems doing so?

Thanks!
 
The images from the X-Pro1 look to have been resampled though - there's a large scale pattern in the noise floor that looks to be results of a small fractional scale.

Graeme
 
You must log in or register to reply here.
Back
Top