12 million is 12 million is still 12 million

Can you estimate the overlap between the two images in brightness space? If the merging is more or less additive ("and" operation) with the proper registration, then you should get more resolving power where the images overlap in brightness space. Regular blending of two images with different exposures is more like an "or" operation so you may not get more resolving power.
--
Leon
http://homepage.mac.com/leonwittwer/landscapes.htm
 
"In the end I don't understand why this has to be re- discussed a zillion times. "

119 replies, Including your long entry (very intelligent even if I do disagree), and you don't think it needs to be discussed? I think for the new S3 owners if nothing else.
 
So Gabe, does my logic from all the white papers seem sound and
correct? It feels right that the S3 in IDR mode is guessing a set
of S and R pixels.

on the mixing of s and r pixels it will keep one of the two, but
not both. lets say for easy thinking, they both have 8 stops of dr.
the S is set to 0-7 and the R is set to 3-9. combined we have 0-9
the 10 stops. but it will use 0-1 from the S and 8-9 from the r.
the mixing is combining the 2-7 from the s and r pix.
Unless I misunderstand what you are saying, that is the right way to do it and I'm sure the one used by Fuji.

Technically it is possible to drop all the overlapping bits between the two sensors. The small sensors would only introduce noise in the mix in the type of light where the big ones performs well. The only information that matters in the small sensors is the one not available in the big one. Fuji is not doing that (photos from small sensors only are pretty good in the whole DR) it is another trick they could have used to save space in the compressed RAW files but they didn't do it.

--
Regards
Gabriele
California, CA
 
You are quite right, you also have a reasonable explanation for the
bump in resolution, the effect of dynamic range on perception.
Thanks for the even tempered response!
I'm always tempered with tempered people. I'm profoundly
un-tempered with unreasonable and dishonest people.
That's how you responded to a new poster here tonite, and his very
first post.

What's the word for 'arrogant' in Italian?
It's "DillonJames".

It takes an arrogant pwerson to write such a sig line:
This info has been brought to you by a 100% genuine S3 user/owner.
Beware of substitutes!
Ghosts ghosts go away, please don't come back another day...
DJ, get some help. The ghosts are only in your head.

You are simply picking up fights & making a personal crusade on people who disagree with you.

DJ, your own venom, the one you spit out here at dpreview; will kill you. I'm serious: you're needlessly littering your mind & well-being (as well as this forum).

--

'Does Phil live on this planet, or is he on his way here?'
(subtitle line in obscure 70s Indian movie :)
 
Yes the increment of resolution is between overlap image in the mid tones this happen between -3.5EV and +3.5EV in RAW images. :)
I’m trying to tell this in my messages. :)
 
I try a lot of mixings in PS the most proximal of S3 Results is blending like used In Sinar Dual Scan Backs. Is impressive the similarities of S3 Files w/ Dual Scan blending, I post one test in some place of this post LOL. :)
 
I Agree, S3 show more details over 1D Mark II (A had one in past), 20D and all 6MP cameras, but show less details comparable w/ D2X (I use both cameras).
 
In-camera interpolation is marketing bs of the worst sort.

Why would you want to make the images bigger in camera, without increasing the actual resolution. It gives you bigger files, which slows transfer down and fills your card faster.
It can all be done just as well in photoshop afterwards!!!
 
Hi Leo,

That files looks great. Later tonite I'll play with it. I also looked at the link you sent me, the pdf, and I have to say, I think it is somewhat misleading of Fuji to use the phrase, "12 MP of resolution."

Question. Your D2X vs the S3. Up to what size print do you see no difference in resolution?

There is a new poster here, AnthonyNJ, who uses a Canon 1Ds2 in his portrait studio. He recently commented to me, that up to A3 sized print, he sees no significant difference in details rendered. What has been your experience?
I Agree, S3 show more details over 1D Mark II (A had one in past),
20D and all 6MP cameras, but show less details comparable w/ D2X (I
use both cameras).
--

This info has been brought to you by a 100% genuine S3 user/owner. Accept no substitutes!

Ghosts ghosts go away, please don't come back another day...
 
If you don't see the ISO readily available
on the LCD or in the viewfinder, we share the same beef.
I dont see it readily available in the viewfinder, and that never bothered me. I set it and as I have got a thing called brain, I can remember what I set it to for the next shots. If I dont remember I look on the small display - I guess the Canons only have one on the backside - and have a look.

Some things are so easy!
 
I then converted the 6MP file in photoshop (bicubic) to 12MP, and the result was rather clear: No additional detail whatsoever was visible in the 12MP pic.
Though I also believe there is no more than 6MP+ for the S2/S3..... All images from the s2/s3 are interpolated to 12mp first. In otherwords, even 6mp images are 12mp that have been downsized back to six in-camera... Not that its a big deal but if there were a difference it might not be seen using this method since the 6mp is used to be 12, but who knows....

Working it the other way around... Many say that downsizing the 12MP image to 6 results in a better 6mp image than other 6mp cameras which is where I think the original "its six but resolves like 8" came from to begin with.

Been awhile since I studied this but I think that is the way it works. Just an FYI for those interested.

Best regards-

Karbo
--
'Great minds talk about ideas, Simple minds talk about people'

http://karywall.com
 
Why would you want to make the images bigger in camera, without increasing the actual resolution.
Because of the sensor shape/layout. Due to this their isnt another way to get the extra info unless fuji were to release an SDK on how to maximize detail based on the octogonal layout. Check some of the posts by Thomas Knoll, where Adobe may have some of this information (or not).

Best regards-

Karbo

--
'Great minds talk about ideas, Simple minds talk about people'

http://karywall.com
 
Hi,
... and everyone who says "the 20d
will beat the S3 hands down" does not know what he is talking about.
I don't own both cameras so my judging is based on Phil tests - both cameras with 50mm f1.5 at f9:

http://www.dpreview.com/reviews/fujifilms3pro/page21.asp

My eyes see a substantial difference in resolution - this is why I said "hands down", not "eyes closed" when I referred to D2x :-)

--
Paulo Abreu,

'It is not worthy to make a video of your life - just keep the best moments in pictures!'
 
I still find it hard to believe that there is a real resolution gain to be had.

The sensor only records 3 million pixels, so any extra pixels has to be created from data thats already there.

Which in turn leads to an 6 mp image that lacks sharpness compared to an image from a proper 6 mp-sensor.

I havent done research on this, I admit, but to me this seems like common sense.
 
I have made comaprison test prints with my Epson 2200, 20D to S3, with the same model lens. The S3 seems to resolve as much detail, in teh prints at least, as the S3. The S3 prints, even in normal dynamic range scenes, to my eye, creates a more filmlike image, especially in highlights. And becasue you can 'expose to the right' more with teh S3, without blowing highlights, I think the shadows and midtones look that much better.

Thom Hogan felt teh 20D actually resolved more than the 20D. And M. Reichmann seemd to think so to.

Dave Eckells from Imaging Resources is on record saying , (a post here) that the 6MP S2 printed up at what looked to him to be about 8MP.

Anders Uschold, from British Journal of Photography, does some of the most technical tests in Europe. He wrote:

“The Fujifilm FinePix S3 Pro posesses a very progressive sensor design. Instead of any external interpolation method known to my test laboratory, here the interpolated mode provides a real increase of detail information. Using high-end, well-suited lenses, the camera can compete with eight million pixel models of other manufacturers.”
 

Keyboard shortcuts

Back
Top