EOS 1D Image Analysis

[Noel_Carboni]

Well, I missed seeing the announcement the minute it was made, though we had all already heard the (IMO) disappointing news that this new camera was going to be only 4.1 MP. Hardly worth a second look. Yet, who could resist at least LOOKING...

I couldn't. Being no stranger to the digital darkroom, I decided to put the 1D's sample images - from http://www.usa.canon.com/EOS-1D/sample.html - through the paces with the good tools.

1. ISO Noise! Worse... Banding noise!

This thing's got a CCD. Guess what? It's got nasty stuff happening in dark places in medium ISO images. Check this snippet of the ISO 400 motorcyclist shot, in which I pushed the luminance curves up on one of the darker areas... Note the noise, and ESPECIALLY the banding:



2. Pattern inaccuracy!

People have said that the hair detail in the image of the girl is good. Is it? It's arguably a bit better than the D30, pixel for pixel, but it ain't perfect. Canon apparently has its problems when it comes to interpreting imager data in images of things with fine lines... I found a couple of kinds of pattern in there. Note the little light squares with dark centers in these nearest-neighbor enlargements:





3. Upsample test.

One thing I like to do is print digital images large on occasion. I've made some stunning 24 x 36 images from D30 shots. How well will the 1D fare? Here's one area where it appears to shine. Check this 4x upsample of two of the small diamonds in the ring in one of the sample images. Based on what I see here, the 1D will print a sharp, though not quite large format sharp, 24 x 36 image.



4. The Carboni special sharpening treatment.

How well do 1D images stand up to my special upsample-sharpen-downsample treatment for sharpening. Fairly well, I'd say. Check this "before and after" out:



I worry that these things won't be fixable in firmware (or simply won't be fixed by Canon). They didn't fix the little glitches in the D30, after all, and I for one made darn sure Canon knew about them.

Lest I sound like a nay-sayer, the 1D images are about on a par with those delivered by the D30's CMOS (except for that banding noise; here the D30 wins), and with 30% more pixels, I guess the 1D will be be a pretty good camera. Stellar? No.

Perhaps not worth trading in my D30 for just yet.

Your opinions?

-Noel

[ September 26, 2001: Message edited by: Noel Carboni ]

I had high hopes for the new Canon, oh well. Not to change the subject but, have you done this test on the Nikon D1X images to see how they stand up to your tests. This would be good information for those like me who are getting ready to purchase. I have almost completely written off the 1D but are the Nikons images any better and how do they hold up to your tests? This would be great comparison information, if of course you have or are willing to do the tests.

Hi all--so far all the images I've seen from this camera look very very disappointing -- I can't believe that IMHO the D30 images are much better. I DO need faster FPS and AF and can justify the cost of this camera for my type of work -- BUT I am NOT going to pay 6000 USD or so for images that are no better (or marginally noisier) than my D30's.
PLEASE all you guys who tested this camera before the announcement please post a decent pic or two.
On the 1D website the Model is totally washed out. Other sample pic seems shooter needs to go back to "Picture Composition School class 101" to learn how to turn camera
correctly.
For a brand new flagship product I would have thought best possible pix were in order with no expense spared to get them right.
Chher all -r

[Rick Fowler]

I also think that the images that Canon posted as samples are of poor quality. Phil over at dpreview said he had many images that he shot but he can't show them or talk about them so hopefully they are still working on this.

[Noel_Carboni]

Richard,

I did do some testing on D1x images some time ago, and recall finding (to no great surprise) that the disparity between the vertical and horizontal resolution was quite apparent. I do remember the noise level being fairly good, but please don't base a buying decision on my (admittedly poor) memory. If I can find some time I'll do a little more comparitive analysis.

-Noel

[carshooter]

Could the fact that the images we've been shown are all JPEGs have any bearing on their perceived quality?

I personally would like to view a TIFF or a RAW file and make the conclusions from that.

Dean Siracusa

[Alon Nahary]

I share the same filling. With this spec and price, it does not worth it for an amateur like me. I stay with my D30.
Regarding the samples, I find it hard to believe that these are original JPEGs from the camera. First, there is no EXIF info. Second, 856K for 4MP is too low (for the bike picture). The artifacts that you see as banding, looks to me like a JPEG pattern of a heavily compressed picture. I guess that they did that in order to reduce the load on their site.

Alon