I’ve promised this noise and “pixels on target” ISO/noise test for some time and I had the opportunity to get some cameras together to pull it off. With the help of Amish Solanki (http://www.asphoto.ca) who shot the video and drove 30kms to my house TWO times to make this possible. You may recognize his online name of Gujustud. This would be a 1Ds Mk.II vs. 5D test if it wasn’t for him & Jon.

The pre-14 bit cameras we used in this test were the Canon 5d, 20D, 1DMk.II and the former flagship 1Ds Mk.II. 12.8, 8.2, 8.2 and 16.7 megapixels respectively. The 5d was used as the control camera that we tested against as many claim it is the cleanest non-14 bit camera. All shots were taken at the same time with tripods beside each other. There was about 5mm between bodies. The metering and lenses were all the same. We used center weighted average with two 50mm 1.4 lenses. We shot totally in M with AWB. Noise reduction is OFF. Heck even the colour space is the same. Understand that EVERY SINGLE shot here is unique, every posted shot was taken at a specific time.

Adobe Camera raw was used to process the images. DPP was HORRID in rendering a clean image and we show that in the video. No edits/adjustments were applied. Swapping lenses and metering modes we also found zero difference*

Two interesting notes: First is the crux of this test – the 1 series camera meters or captures what appears to be 1/3 or even 1/6 of a stop less. The image is just a touch darker than the 5D control images and as we all know, exposure at higher (or any) ISO is critical to noise control. The Sekonic light meter was of little use other than a control to verify when we had the meters in compliance.

*Secondly, the 1Ds Mk.II (not the 1DMk.II) appeared to have a far more sensitive meter at the higher ISOs. At 3200 we had the meter on the 1Ds Mk.II metering a little lower than the 5D, even EV and sometimes swapping spots on the meter marks. Concerned we did some pretesting shooting by meter and then shooting by static lens settings. “Huh?” you say? Well one experiment led us to set the aperture and exposure exactly the same. Then we tried to just show even on the meter with a differing shutter speed and constant aperture. Results? We found the 1Ds Mk.II meter to be VERY precise and metering a touch darker than ALL other cameras. Some camera’s meters appeared to “warm up” once turned on where the 5d was on the whole time. When it came time to test and actually press the shutter, the meter was the same between both cameras. We got lucky and had perfect light so that the test would have best exposure we could aim for.

Does this mean the 5d has a meter that is calibrated to let the camera control noise better? Is the 1 series best used with external meters and by people that are assumed to be skilled? Utilized best by people who shoot to the right in M or what have you? I won’t get into the semantics of that, the images will show what the cameras do in the hands of someone who will shoot with 0 +-EV and we also shot at +1Ev then pulled back -1EV in post (shooting to the right) to see if there was an advantage there. Be mindful you will see a touch different colour and exposure from the 1 series cameras. If you think this contribute to a bit more noise but a truer shot then so be it. If you think this makes the 5d slay the 1 series cameras for noise control then you can draw that conclusion.

Summary:
My overall summation is that the 1Ds and the 1D make a better looking picture in the right hands. The 5D wins (yes I said it) wins in the noise category but by a contested hair. My wife says “It looks %25 better noise wise, but the other one (1Ds Mk.II) image looks better, more real.” Is the sensor/metering allowing a bit of noise to creep in? The 5d IS NOT blowing the other cameras away in noise. Sorry. It beats (?) them – beats them like the photofinish at the Kentucky Derby and no one can tell until the photo is developed. The 1Ds Mk.II, 1D Mk.II and 20D all look good noise wise and they have their price and feature merits.

These cameras ALL have replacements that you can buy new in stores (5d replacement is coming presently). These are great used buys that will serve you well for years. Keep that in mind.
Well enough pontificating about what I think the following images do. You can decide and think about what conclusion YOU come to. Enjoy the images, the video and let’s hope everyone can take something away from this spirited but nearly scientific test.

————————————

The Test, The Images, The Video

*hi-def video link coming Thursday – check back!*

Above is the hi-def video of us doing the test. Thanks again to Amish Solanki for making the video and supplying the 1D Mk. II. Jon brought his 20D and it impressed as well showing some great detail.
First up is ISO 3200. The 5d is the winner here – BARELY. Some may say the 1Ds Mk.II wins here. The 1Ds Mk.II is showing a bit darker and that contributes to the noise. My opinion is that the image clarity when seen on the full screen tips the hat to the 1Ds with an ever so slight gain in noise. Barely. This is NOT the 5D slaying the 1Ds Mk. II as some would have you believe.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

Ok, how about shooting to the right @ 3200? Screw lens and aperture, real world shooting. Let the meter read +1 EV, shoot then bring it back -1EV in post processing. Let’s see…well the 1Ds wins here with the detail. The noise difference is negligible. This is where the 16.7mp comes into play. Maybe there was tripod wobble from a small earthquake but they were taken at the exact same time on sturdy tripods with a shutter speed that was FAR above focal length of 50mm. No explaining this one away. I popped them back in forth in hopes of seeing if the is blur but the tripod and the focus ring never moved.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

How does the 5D stack up against the 1D at 3200? The 5d seems to just have a larger 1D sensor. 5D wins here. They’re the same price used, more resolution from the 5D but weaker AF, features etc. Let your wallet and you need for speed decide.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

Ok so you can get a 5D on the cheap. What about the now $500 20D which is the same megapixels as the 1D Mk. II? How does it handle ISO 3200 against the 5D? Crop factor comes into play here. Does it pull in more detail? Have a look.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

Well there is the ISO 3200 game. It’s a tough racket trying to make everyone happy with metering, details and noise suppression. Does EVERYONE shoot at ISO 3200? Hell no. So let’s move on to some realistic ISO settings. ISO 3200 is a software generated setting from 1600. ISO 1600 is the true highest ISO rating that these sensors do without invoking the H setting. So let’s see what happens there.
5D vs 1Ds Mk.II again. Round two. 5D wins on noise again by the slimmest hair. 1Ds has the details and the better colour. What to say? You may think otherwise.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

5D vs 1D Mk.II? **EDIT**. Some discussion ensued on POTN. I was previewing these using Vista’s image viewer. I was clicking back and forth quickly to evaluate them then I wrote text and uploaded.

The 1D Mk.II image won. (this is past tense)

BUT – I look at them in zoombrowser and look at them with Firefox and the 5d looks a touch better. I open them again in ACR at %200 and the 1D looks better – slightly. The overall image looks better,but the noise is slightly more apparent on the 1D Mk.II. Colour rendition seems to add the depth here to the fact the internal processing is a touch different.

Be mindful everyone when pixel peeping that you are looking at how a program INTERPRETS the pixels.

Both shots taken at the exact same time. Metering checked with Sekonic L-358

Here comes the underdog again, the 20D. Does it do much in the noise department at 1600? I call this a dead heat, resolution balanced by cheap!

Both shots taken at the exact same time. Metering checked with Sekonic L-358

Now we head into enemy territory. ISO 400 and lower.Barbed wire, land mines and fanboy death squads roam the desolate landscape. The clean ISO ranges. As we got down here I sacrificed shutter speed to get a reasonable DOF. I assumed that the tripod would be my rock. There are two shots here between ISO 400 and 100 that are a bit OOF from camera movement I assume. Noise is always noise, but the detail is fuzzy on two. Apologies but time constraints dictate I can’t spend the entire weekend pixel peeping with a wife and needy kitten orbiting. Plus most people are here to see the 3200 stuff I’d been promising.

Images are labelled accordingly for your viewing pleasure.

Matching shots taken at the exact same time. Metering, again checked with Sekonic L-358

pixel peeping leads to fear. Fear leads to anger. Anger leads to hate. Hate leads to suffering. I sense much fear in you. – Lord Malone