• Please use real names.

    Greetings to all who have registered to OPF and those guests taking a look around. Please use real names. Registrations with fictitious names will not be processed. REAL NAMES ONLY will be processed

    Firstname Lastname

    Register

    We are a courteous and supportive community. No need to hide behind an alia. If you have a genuine need for privacy/secrecy then let me know!
  • Welcome to the new site. Here's a thread about the update where you can post your feedback, ask questions or spot those nasty bugs!

Sample image EOS-1Ds Mark III

Hi folks,

I frequently see requests for sample images of the 1Ds3 on various internet fora. People want to see for themselves if the image quality suits their needs. Therefore, for a limited time (due to hosting space limitations), I'll put up a full-size sample JPEG image at this URL (note: 21MB !):
http://www.xs4all.nl/~jlvdwolf/temp/7640_CO40_FM1-175pct_sRGB.jpg

The image was shot on tripod, with mirror lock-up, with a TS-E 90mm lens at f/7.1 (at the onset of visible diffraction) at ISO 100. It was manually focused (with LifeView) on the name shield of the windmill. Composition was not taken into consideration, it's just a test-shot containing several material structures and sky. I don't claim any artistic qualities (in fact composition is poor) nor were such qualities intended (to reduce the risk of commercial exploitation).

Raw conversion was done with Capture One version 4.0 with the default camera profile, and the default 'film-curve'. No noise reduction and no sharpening was applied at that stage.
The resulting image was only sharpened once with Focus Magic in Photoshop and, after adding the title, converted to 8-b/ch sRGB and saved as a maximum quality JPEG. Besides the Raw conversion, I applied neither tonemapping nor dust removal (2 dust spots are visible) whatsoever, so it's basically a straight Raw conversion plus a sharpening step in Photoshop.

I hope that it will help to get an idea of what's possible.

Enjoy,
Bart
 

John Sheehy

New member
I'll put up a full-size sample JPEG image at this URL

I could give a detailed IQ analysis, but the first thing that comes to mind, is "WOW".

Anyone who says 6MP is the practical limt, that the 1Ds has no more resolution than the 5D, etc, seriously needs to get their eyes checked! This image is flirting with aliasing at the pixel level (but no tetris pieces are seen). Good show of the lens, as well.

There is no doubt in my mind that this camera has the best IQ of any DSLR. The D3 has a hair less image-level shot noise, but that's the only way in which it betters the 1Dsmk3, and probably by an inappreciable amount.
 
@Ralph and Nikolai, you're welcome, and as you can imagine I'm overall pretty pleased with the camera myself.

I could give a detailed IQ analysis, but the first thing that comes to mind, is "WOW".

Thanks, it is the same reaction (for each and every image that can be taken with the right technique) that made me post it. Initially I was struck by some of the poor results that were posted when the 1Ds3 came out, but it apparently proves to be technique related, the camera is not to blame.

Anyone who says 6MP is the practical limt, that the 1Ds has no more resolution than the 5D, etc, seriously needs to get their eyes checked! This image is flirting with aliasing at the pixel level (but no tetris pieces are seen). Good show of the lens, as well.

I knew you'd appreciate some of the technical subtleties, both of us we're not that different when we look at potentially distracting 'features'. Even an artistically sound image can be ruined by technical imperfections in our eyes. And it's just like looking at a pimple in a mirror, it seems to grow with prolonged viewing.

As for the 'flirting with aliasing' (I like that characterization), I did adjust my regular post processing technique a bit for this image (it differs a bit, depending on actual image content). The fact that the sensel pitch is smaller than I was accustomed to with my 1Ds2 will be more taxing for the lenses (their inherent resolution is sub-sampled a bit more). The combination of lens (in this case at the brink of viible distraction limitation), anti-aliasing-filter, and sensel geometry, results in a lower MTF response at the pixel level. This requires an adjustment in the high spatial frequency sharpening one applies.

When using 'Focus Magic', which deconvolves the image content, I typically have to increase the 'amount' setting compared to larger sensel pitch cameras (the 5D might be an exception due to its mild AA-filter). In addition to increasing the 'amount' to 175% (as indicated in the filename), for this image I even left out the adaptive blending layer approach that I normally need to use to reduce halo and stairstepping artifacts, when pushing the limits. Again, this is image dependent.

There is no doubt in my mind that this camera has the best IQ of any DSLR. The D3 has a hair less image-level shot noise, but that's the only way in which it betters the 1Dsmk3, and probably by an inappreciable amount.

I agree, although the slightly higher native sensitivity and much lower resolution of the D3, make the comparison, well, let's say difficult. I do have to mention that the relatively small sensels (and associated/new camera electronics) of the 1Ds3 do limit the dynamic range of the 1Ds3 to =~ 11.3 stops (according to my own analysis). This is indeed confirmed by the many practical (bracketed) shots I have taken of various subjects. Slight (1/3rd stop) under-exposure effectively produces visibly more noise and (visibly a bit more testing) reduced dynamic range in 1Ds3 images. Bracketing is something I use much more often with the 1Ds3 than with larger sensel pitch cameras before. I do think that, unless there is a technological breakthrough in both 'read noise' and 'well depth' and/or 'gain level', resulting in higher dynamic range per pixel, I won't invest in a smaller sensel pitch camera than what the 1Ds3 offers.

A deciding factor in getting the best resolution out of 1Ds3 images so far, also is the use of Phase One's Capture One (in particular version 4.x). No other Raw converter, that I've tested sofar, is capable of extracting so much resolution from the 1Ds3 Raws. However, it does also boost noise, easily mistaken for detail, so one is required to limit fundamental noise as much as feasible, before optional noise reduction is even considered.

Bart
 

Bob Ring

New member
Moire in image

This is a great image showing the capability of the 1DsIII (which I own as well). I use Phase One Capture 3.7 vs the ver 4 (perhaps I'll have to try it!). The problem I'm having with the 1DsIII is moire. When I look at this image I see it on the bricks as a wavy pattern If I zoom in it goes away. I took a photo of a model with a hat that had a band that showed the same thing. I had to select the area and apply a gaussian blur to it. Wondering if anyone else has had/seen something like this and how you solved it. Appreciate any comments. Regards, Bob Ring
 
The problem I'm having with the 1DsIII is moire. When I look at this image I see it on the bricks as a wavy pattern If I zoom in it goes away.

Depending on which application you use to view the images, you are probably seeing moiré downsampling artifacts at any zoom setting below 100% (actual size). Most image viewing/editing applications use a quick-and-dirty method of downsampling, which leads to quick-and-dirty results.

This is not specific to the 1Ds3 or any other camera, but it's basic digital signal processing which states that resampling requires pre-filtering to avoid artifacts. The same thing happens with film scans, although it manifests itself in more complex shape due to the presence of graininess.

Bart
 

Bob Ring

New member
Moire...

Thanks Bart. I'm not sure I follow you though. I looked at your image here simply in Explorer and my image was viewed in Photoshop CS3 of the hat band. Two different scenarios, right? Thanks.
 
Thanks Bart. I'm not sure I follow you though. I looked at your image here simply in Explorer and my image was viewed in Photoshop CS3 of the hat band. Two different scenarios, right? Thanks.

Sure, two different scenarios. However, Internet Explorer will resample the image down to fit it in a window. Due to the fast, but technically incorrect, method used it will result in artifacts (even if the image itself has none.). You mentioned that zooming in made the moiré go away, which is consistent with what I described when you reach the 100% zoom setting. When you see moiré at 100% or more, I suspect your (LCD?) monitor may need a little tuning of the moiré setting.

In my experience the 1Ds3 is not overly sensitive to aliasing, but it is still possible to get it under certain scenarios. That type of capture moiré will be visible in the image, even at 100% zoom. If the image setup allows, you can use an aperture that's narrower than f/7.1 . Such narrow apertures will cause increasing amounts of diffraction as the aperture gets narrower, and that will lead to a certain softening (low-pass filtering) of the image before it reaches the sensor array. That will effectively reduce the aliasing that originates from the discrete (under-)sampling process.

It would be interesting to see a 100% zoom crop of your hat band moiré since, as I said, it is not that common on a 1Ds3. It would also be interesting, if you still have access to that hat, to try and eliminate the moiré by using progressively smaller apertures (or closer shooting distances, or minor de-focus), and get a feeling how much can be prevented at the time of shooting. An application like "Focus Magic" will allow to recover at least some of the reduced sharpness.

Bart
 

Bob Ring

New member
Thanks Bart. Unfortunately, I don't have access to the model and the hat but I do have the photo of course. I also have a photo of an old building that Moire shows up on the roof. I'm traveling now but will try to post (or link) them here for you to see later this week. I'd appreciate any help on this as it seems to me that moire is an issue for me but perhaps it's something I'm doing in the processing of the raw file. Thanks, Bob Ring
 
Bart, here is a jpeg of the model with hat band. Not sure if you'll see the moire or not on your screen (http://bobring.zenfolio.com/p227892081/?photo=179264905).

Thanks Bob,

Yes, it's clearly visible. Is that a 100% zoom (actual size) crop, or was it resized?
Would you mind sharing with which lens and at which aperture the shot was taken?

I'm not sure how to show off the moire on the building's roof shot. I see the moire in my print (full tif file) and only sometimes on the screen!

Hmm, seems something else is interfering there. The screen (if LCD) is more likely to show it.
Maybe there is some sort of resampling used in the printing process?

Bart
 

Bob Ring

New member
Bart, the camera/lens info: 1DsIII with 70-200mm f4 @ f8 with flash. Phase One with sharpening in raw (amount of ~136 and radius of 2 in soft look mode) processing into CS3. Image is 300px/in with a sz about 10" wide by 13.4" tall. The crop is 300 px/in and about 5.8" by 3" tall. I can see the moire right in Phase One (sames with the roof of a bldg photo not attached here).
 

Bob Ring

New member
More moire (in roof)

Bart, here is a 2Mb JPEG of the bldg photo I've mentioned before with the roof showing moire: http://bobring.zenfolio.com/img/v2/p977842027.jpg
The camera/lens conditions are: 1DsIII with Canon 24-70mm lens (48mm) at f10. Again, moire is really strong in a tif file print of this image and is visible at different magnifcations on my LCD screen. I'd love to be able to avoid this in the future and/or know how to correct it after the fact. Thanks, Bob Ring
 

John Sheehy

New member
Bart, here is a 2Mb JPEG of the bldg photo I've mentioned before with the roof showing moire: http://bobring.zenfolio.com/img/v2/p977842027.jpg
The camera/lens conditions are: 1DsIII with Canon 24-70mm lens (48mm) at f10. Again, moire is really strong in a tif file print of this image and is visible at different magnifcations on my LCD screen. I'd love to be able to avoid this in the future and/or know how to correct it after the fact.

Bob, you still haven't answered the most important question, AFAICT. What are we looking at here; a crop of the original pixels, or something else? If it is the latter, the examples are, unfortunately, meaningless. Resizing an image can completely alter the pixel characteristics of an image. Any camera and lens can alias with bad downsizing.
 

Bob Ring

New member
Moire question

What is AFAICT? How would I show you the full sz? Yes these (model and band and building) are not full sz. I was trying to be able show what I see on my LCD monitor And when printing on my Epson 7800 what I see when I look at the full sz tif file. Moire visible on both LCD & Print. If you can tell me the best way to show a full file I'd be happy to. Thanks.
 

Cem_Usakligil

Well-known member
What is AFAICT? ...

If you can tell me the best way to show a full file I'd be happy to. Thanks.
Hi Bob,

AFAICT: As far as I can tell...

Open the original file in your image processing prgram. Crop the image to a small area which includes the moire, say about some 400x400 pixels. Save that as a jpg file without sharpening or resizing. Post the crop area jpg.

HTH
 
I should have known AFAICT! Anyway, here's a crop taken from the full sz. image& saved as a JPEG 450x450 px w/no sharpening or resizing: http://bobring.zenfolio.com/img/v2/p6863706.jpg
Thanks.

Hi Bob,

The roof in that crop does look a bit strange, but it isn't caused by moiré IMO. It looks a bit more like residual lens aberrations, although that's not that likely given the lens and aperture you used. Maybe there was a bit of camera shake/rotation, I don't know the shooting details but the EF 24-70mm can be very sharp around f/8. Stopping down to narrower apertures than f/7.1 will (on the 1Ds3) reduce aliasing due to the introduction of diffraction blur at the pixel level. That will function like an additional low-pass filter at capture time.

In the earlier (downsized) copy of the full scene of this building there was some aliasing, but that was probably caused by the downsizing algorithm.

Bart
 

John Sheehy

New member
In the earlier (downsized) copy of the full scene of this building there was some aliasing, but that was probably caused by the downsizing algorithm.

It looks like nearest neighbor or similar at play. Not only does NN lead to unecessary aliasing, but it also increases image-level noise, sometimes tremendously. NN and similar downsizing methods are a big part of the problem with the anti-MP crowd; the more you need to downsize an image to fit it on the screen, the more the displayed image noise increases.
 
Top