SD9 & dRebel XT - two outdoor same lens tests

..you see the entire back end of a car in the Rebel shot that is missing in the SD image.

Now this is probably a function of sensor FOV, but when your seeing car sized discrepencies in image framing -- shouldn't you adjust for it?

(if the framing were the other way around here -- there would be howling!)

If you were comparing a full sized sensor with that of a 1.7, would you use the same methodology?

--

Peace.

========================
http://www.pbase.com/jarvic7/root

 
In terms of speed, The camera itself is fast, I think he is referring to computer bandwith issues that would be introduced by larger images.

Not a concern of mine.

(I have it on good authority that computers will probably get faster and cheaper as time goes on ;-)
--

Peace.

========================
http://www.pbase.com/jarvic7/root

 
This is the first time I've sseen comparisons of the SD9 with a DR, using the same lens, with parameters stated. On the other thread, pictures taken indoor. This time, it's outdoor photos.

Are you lying again ? Or this is sense of humor ?
However,

This has been done over and over and over and over. I really don't see
what new light would this test would bring.

It's like beating a dead horse with a different whip.

except in this case a different alias.

--
Chunsum.

http://photography.chunsum.com
http://www.pbase.com/chunsum
http://www.pbase.com/sigmasd9
http://www.foveonx3.org
 
goert:

This thread will prove to be a popular one -- and it was clearly labeled, so it was easy to skip.

popular threads are popular -- for the simple reason that they are popular!

Why get upset about it? Just move on to a topic of interest, or create one of your own.

And actually, this topic would be enormously interesting to someone who was considering both cameras (and do realize that for everyone who posts messages here, there are probably 100+ who are silently crusing this forum for topics just like this one ).

Lighten Up Buddy!

--

Peace.

========================
http://www.pbase.com/jarvic7/root

 
Actually, both cameras use unique x/y locations for each
pixel/sensor in each color channel. The practical difference is
the DR's sites are smaller, which allows them to be aligned/wedged
next to one another. But the x/y resolutions in each color channel
doesn't get higher due to interweaving them.
This is incorrect. The foveon has 3.whatever m physical photosite
locations with differnt x/y locations wheras the DR has 8.whatever
m phyiscal photosite locations with different x/y locations.
It isn't intuitive, but it is correct. The three channels sample from mutually exclusive populations, so it doesn't matter if samples topologically overlap or not. All that matters, from a resolution POV, is the number of sensors in each individual RGB exposure which are then embedded in the final composite RGB image. In fact, when you print, the printers pixels/sub-pixels might not be aligned the same way a monitor aligns them.
The bayer sensors obviously see only one color at each site but use
neighboring sites to interpolate the other colors as well as detail.

If the foveon would have the same number of physical site locations
it would trounce the DR in detail.
It's already winning in terms of color resolution, with 3M color to 8M monochrome photosites.
On a different nore, the smaller sites certinaly don't seem to hurt
the DRs noise levles, but I think the SD9 does show a little more
dynamic range.
Well that seems to be one of the issues at foveon that they need to
work on. The Polaroid sensor showed that even more.
I think that's about the right number. I appreciate the lower
bandwidth, if nothing else, it makes everything inthe system run
significantly faster and is a lot more storage friendly. After
5MP-X3, I'd like to see them stop increasing MPs and switch to
concentrating on pixel quality.
Well bandwidth is actually a problem. The foveon needs a high
bandwidth because it is 3 times the information (if compared to the
same number of physical site locations) and there is no dedicated
hardware to do a lot of the processing as there is for bayer
sensors.
Internal to the camera that is true, it's simply working with 10M data point RAW files and they have to be produced and swallowed by the CF card. But once the data is assembled into a picture, the same amount of optical data is carried in far fewer pixels, as we see in the demo images.
The extra bandwidth also affects image playback and examination
speed, where the DR is always 2-3 steps/2-3 seconds behind your
fingers. The SD9 is very quick on its feet by comparison. I think
that is important, image review is the essence of digital.
I havent played with the rebel but i think this is more a limit
imposed by canon on the low cost body. Higher priced bodies are
very fast.
Actually I think it is more a function of Canon slowly improving the computational speeds of their cameras. The original 1Ds was and is very expenisve and it was also very slow to operate in review mode compared to the XT, assuming you count no magnification/panning as a comparable review mode.

The 10D was even slower, a lot like a P&S from the day. The 1DMK2 is a lot slower than 20D/XT when reviewing uncached images, and all of these cameras are very slow compared to the circa-2002 SD9. The reason for that is undoubtably fewer recorded pixels to manage, not electronic slowness in some absolute sense.

Image to image timing data below. Magnifying/panning often present a larger speed disparity, though not originally offered on the very expensive Canon 1 bodies:

http://www.dpreview.com/reviews/canoneos1dmkii/page12.asp
http://www.dpreview.com/reviews/canoneos1ds/page11.asp
http://www.dpreview.com/reviews/canoneos350d/page12.asp
http://www.dpreview.com/reviews/canoneos350d/page12.asp
http://www.dpreview.com/reviews/sigmasd9/page10.asp
As for the body, the SD9 is too different than the DR to compare.
It was a $2000 MSRP price point DSLR in 2002 dollars. That's about
where it's build quality fits in today, very similar to an S2 Pro
body.
Obviously the rebel is a body meant for a lower price bracket.
Very much so.
 
Mobsie,

Do you own this forum or something ? Or do you own the Foveon
technology ? Let us know which one. Cause you sure act like one.
I dont, Trollie.
Why don't you go and start a new thread specifically for childish
name callings. I bet you with many of your persona could show off
some wild creativity.
Now, giit, Mobsie. Hurry out the here.
Do you own this forum or something ? You sure act like it.

Trollie, I am not the one with many persona here, and btw, who are you?

--
--
Cheers
Günter

SD10-images website:
http://www.pbase.com/ghoerdt

Panorama website:
http://www.pbase.com/ghoerdt/panorama
 
Mobsie, what do you want ? Stirring troubles again ? Haven't you had enough fun yesterday and all morning today ?
Trollie, I am not the one with many persona here, and btw, who are
you?

--
--
Cheers
Günter
 
that gives it a slant towards the SD of course and underlines the
rebels relative resolving power some more

all in all it isnt huge though but it would have made sense to
correct by taking a little step forward with the DR
Yes you are right. But if you don't accept lower resolution in exchange for wider FOV, you've paid $8000 for a full frame camera just get bad corners.

I agree that the DR FOV isn't much different, but I think lens-constant resolving power should be measured, since FOV is a known in any case. Everyone understands that changable lenses can trade one the other.

For example, the 1Ds should be rated as a fairly low resolution camera with a very wide FOV. The SD9, much the opposite.
 
This is the first time I've sseen comparisons of the SD9 with a DR,
using the same lens, with parameters stated. On the other thread,
pictures taken indoor. This time, it's outdoor photos.
I'm certain it's never been done using the same lens. Nor have I seen same-lens resolving power measured without moving one of the cameras closer than the other.
Are you lying again ? Or this is sense of humor ?
I think he may have misunderstood what's being done.
 
dont know what FF cameras have to do with this.

if you want to compare the resolving power of one sensor to another you need to keep the FOV the same otherwise of course you are favoring one sensor

In real life you would generally do the same thing. You frame a specific subject.

--
Michael Salzlechner
http://www.PalmsWestPhoto.com
 
true

i was surprised that they actually released it. It was pretty bad
publicity as well.
The best it could've done was help classify Foveon as lower end. From a purely selfish POV, I'm happy it didn't make it.

Realistically though, it's hard to imagine Foveon wedging themselves into enough high end units to make it work. They clearly underestimated or misunderstood the cartel business culture of the key players.
 
dont know what FF cameras have to do with this.

if you want to compare the resolving power of one sensor to another
you need to keep the FOV the same otherwise of course you are
favoring one sensor
I disagree, the only way to measure a sensor's resolving power is by using the same lens from the same distance. If you put a zoom lens on one but not the other, your helping that one, same goes for moving it closer. FOV differences on top of measured resolving power are easy for anyone to account for.

I knew this would come up, and both camps have valid points, but fortunately in this case 1.6 vs 1.7 isn't much to worry about.
In real life you would generally do the same thing. You frame a
specific subject.
I wouldn't buy a FF camera to frame the subject the same way. At that point all you've bought was poor corners.
 
This is the first time I've sseen comparisons of the SD9 with a DR,
using the same lens, with parameters stated. On the other thread,
pictures taken indoor. This time, it's outdoor photos.
I'm certain it's never been done using the same lens. Nor have I
seen same-lens resolving power measured without moving one of the
cameras closer than the other.
Are you lying again ? Or this is sense of humor ?
I think he may have misunderstood what's being done.
OK, so you compare the SD9 and the DR using the same lens. what really is the point behind it? I really do not see what is there to accomplish here by doing so. Are you testing the lens? Are you comparing the SD9 and DR? is this the exact same lens? why not compare them with the Sigma 50mm? Why the DR and not the D20? what about the D70 or the S3? and finally what importance does this test have? How will it's results affect the rest of the world?

Please, enlighten me.

--
Chunsum.

http://photography.chunsum.com
http://www.pbase.com/chunsum
http://www.pbase.com/sigmasd9
http://www.foveonx3.org
 
It isn't intuitive, but it is correct. The three channels sample
from mutually exclusive populations, so it doesn't matter if
samples topologically overlap or not. All that matters, from a
resolution POV, is the number of sensors in each individual RGB
exposure which are then embedded in the final composite RGB image.
In fact, when you print, the printers pixels/sub-pixels might not
be aligned the same way a monitor aligns them.
No. As we all know it is not correct but if you want to believe this go right ahead
It's already winning in terms of color resolution, with 3M color to
8M monochrome photosites.
As i said it is in some areas resolving less and in some a little more. That even though the FOV was favoring the SD image. At the same time you get digital artifacts on the SD that you dont get on the bayer
Actually I think it is more a function of Canon slowly improving
the computational speeds of their cameras. The original 1Ds was
and is very expenisve and it was also very slow to operate in
review mode compared to the XT, assuming you count no
magnification/panning as a comparable review mode.
No. The higher end cameras always had higher processing speeds. It is simply a matter of money and marketing. The lower end cameras have to have something less to a) cost less and b) not cut too much into higher end bodies sales.
The 10D was even slower, a lot like a P&S from the day. The 1DMK2
is a lot slower than 20D/XT when reviewing uncached images, and all
of these cameras are very slow compared to the circa-2002 SD9. The
reason for that is undoubtably fewer recorded pixels to manage, not
electronic slowness in some absolute sense.
The SD has nice image to image review time but at the same time it requires a full 10 seconds to write an image to a CF card where the canon will write it in just over 1 second. Most timings on the SD are pretty slow.

it takes the SD9 more than double time to show a preview after taking a picture for example. So if you want to look at pictures taken previously the SD9 may be fast but if you want to take pictures or look at the last one taken it is slow.

But that all wasnt supposed to be part of the discussion. This discussion should have been simply about the difference between a bayer sensor and a foveon sensor.

--
Michael Salzlechner
http://www.PalmsWestPhoto.com
 

Keyboard shortcuts

Back
Top