Topic: H39MKII vs Canon 5DMKII image crop comparison (Read 17342 times)

neil snape · « **on:** February 09, 2009, 10:48:44 am »

As promised here is a little crop of an image shot with the same light, similar f stop around f11 and similar enough almost default development in their respective applications.

The Canon is new to me. I don't know it well yet and find it focuses on the nearest zone in the AF points rather than having a central weighting for focus which intended or not the older 5D does.
I found the Phocus application easy to learn but only had it for a few days. The defaults are acceptable but there could be some room for improvement in their defaults settings.

The Hasselblad was shot on the 120 macro, at ISO 100, the Canon too. Development was very slightly adjusted to make the image look close enough to one another on the default settings, and reduced the sharp radius to 0.5 pixels. Increased the moire (or anti moire) in Phocus to 2.

They are crops from 16 bit images saved in jpg at 12 quality. Nothing conclusive but if you open them in Photoshop you can see there are differences.
overview
Canon5DMKII
Hasselblad 39 MKII
Canon jeans
H39MKII jeans

francois · « **Reply #1 on:** February 09, 2009, 11:02:36 am »

The correct link for the H39MK2 file is: http://www.neilsnape.com/canon5DvsH39/H39MKII.jpg

Dustbak · « **Reply #2 on:** February 09, 2009, 11:40:53 am »

Why is the H3D image looking so bad? It is if it is OOF?? Also strange artefacts on her mouth and in her neck but I guess that is part of the reflection on the make-up?

Carsten W · « **Reply #3 on:** February 09, 2009, 11:53:00 am »

Quote from: Dustbak

Why is the H3D image looking so bad? It is if it is OOF?? Also strange artefacts on her mouth and in her neck but I guess that is part of the reflection on the make-up?

There are some very ugly artifacts in both images along the hair-face area on the left side, and also in the hairline above the face in the 5D2 image. It looks like posterization. What kind of processing was done? Additionally, the 5D2 image was cropped a little closer, possibly adding to an impression of greater detail. The 5D2 image looks more tense, and the Hasselblad image looks a lot more relaxed, but the conversion to JPG left some critical artifacts, distorting the final impression. Were the colour spaces not set correctly or not converted to sRGB for web viewing?

Dustbak · « **Reply #4 on:** February 09, 2009, 12:00:39 pm »

Why the moire reduction, did you see moire with the slider set to 0?

neil snape · « **Reply #5 on:** February 09, 2009, 12:12:17 pm »

There are way too many rainbows, which is not the same as moire. Yet with Phocus you need to set the moire filtre fairly high. I didn't crop the jeans in the pictures but the points of contrast with the Hasselblad have coloration all over the place.
BTW I did a still life with a H39MK I and a 800 mm the day before and it also showed the coloured points. I cannot say if in FlexColor if this is a problem.

The images are in Adobe RGB , the Canon converted from ProPhoto from LR.

The Canon has the typical red shadow transition problem , less than the 5D MK 1 but neverless still there. Hasselblad is much better in this respect.

However the , either Phocus or the camera are noisy. I am glad you saw the strange things happening , and no it's not makeup. It is not in the Canon images, nor was it to my eye.

The crops were done free hand so there are pixel level crops. Obviously the Canon has a lower pixel count. Looking at images , well you should be looking at 100% in Photoshop. If you do you should see the images have an Adobe RGB space embedded.
JPG compression could add some posterisation but being nº 12 quality not much.

For the settings read the first post.

Dustbak · « **Reply #6 on:** February 09, 2009, 12:23:13 pm »

I opened in PS. I did some sharpening to the H3D image. Even with some sharpening it doesn't look like the thing I am used to. The artefacts are weird, maybe a result of the moire reduction (but I am guessing).

It doesn't look too good to me. Would you be willing to provide the raw file via posting a yousendit link?

neil snape · « **Reply #7 on:** February 09, 2009, 12:47:21 pm »

Here's another process, with DPP 3.5.2.

A bit too saturated for me but there is a certain softness that would take out a lot of retouching.
Canon with DPP

Wayne Fox · « **Reply #8 on:** February 09, 2009, 12:48:53 pm »

Guess I'm confused. If the original crops were the same FoV, shouldn't the Hasselblad file show considerably less face than the 5D file if I'm viewing original files at 100%? The eyes in that file are smaller than the eyes in the 5DMk2 file.

neil snape · « **Reply #9 on:** February 09, 2009, 01:10:39 pm »

Quote from: Wayne Fox

Guess I'm confused. If the original crops were the same FoV, shouldn't the Hasselblad file show considerably less face than the 5D file if I'm viewing original files at 100%? The eyes in that file are smaller than the eyes in the 5DMk2 file.

Hmmmm, The originals show the Hassy slightly larger than the Canon. The framing with the Canon was tighter though. The file size are correct in full size> 120 MB Canon and 220 MB at 16 bit. Yet , 7000 pixels compared to the 3400 of Canon should be a bigger difference agreed.

Panopeeper · « **Reply #10 on:** February 09, 2009, 01:59:35 pm »

Quote from: neil snape

The file size are correct in full size> 120 MB Canon and 220 MB at 16 bit. Yet , 7000 pixels compared to the 3400 of Canon should be a bigger difference agreed.

The file sizes are reflecting the pixel counts x 6; there is nothing wrong there.

Wayne Fox · « **Reply #11 on:** February 09, 2009, 05:22:57 pm »

Quote from: neil snape

The framing with the Canon was tighter though.

Ah ... yes seeing the full frame it appears the Canon FoV is about 20% closer than the Hasselblad.

FWIW, I used to shoot my Hasselblad/P45+ for portraits until I bought the 1DsMark3. I much prefer the look of the Canon for portraits except large groups where small detail may be important. Not sure if the anti-aliasing filter just works well to improve skin or what. Appears the 5DMk2 should be a great choice for portrait work.

lisa_r · « **Reply #12 on:** February 09, 2009, 06:01:01 pm »

Hi Neil, have you been using the Hassy for a while? How do you feel about the output from the respective cameras?
Also, your link to the DPP output does not seem to work...

Thanks.

neil snape · « **Reply #13 on:** February 10, 2009, 01:51:38 am »

Quote from: lisa_r

Hi Neil, have you been using the Hassy for a while? How do you feel about the output from the respective cameras?
Also, your link to the DPP output does not seem to work...

Thanks.

Yes I wish they didn't have the http auto write in the link tag as copying over links one has to be careful. I fixed it.
As I said before I had the Canon on the day of the shoot and the Hassy both the 39MKI and II two days before.
Phocus software is very well done, and very easy to understand. A few places need some changes but nothing drastic.
I find the beauty in the camera itself, as the Canon is so cheap in comparison. The Hassy has more detail, less color artifacts in most areas, but the rainbows around contrast points are not good. I think this is a software de-mosaic fault but should not have slipped through. IT is on both the Hasselblad 39s I tried two different lens, two different computers.
The Canon for the price is a fair deal. For any pictures where smaller prints are okay , in fact the 5D outperforms the 5DII in speed, focus. The new one does however have what it takes to get into the big league. The Nikon 3x though if the price was lower would have been the better choice. The Hasselblad cameras are very reasonably priced.

Dustbak · « **Reply #14 on:** February 10, 2009, 03:10:37 am »

Have you tried processing the Hasselblad files through Flexcolor? Just to see if this rainbow effect is due to the raw converter or the camera(s) itself. I have not seen this effect myself (CF39 & CF39MS) with files unless we are talking about moire which can be invoked fairly easily on many fabrics.

I like Phocus as well but do see several points that need improvement, it is still an ongoing development in those areas. Quite frankly I still prefer Flexcolor for the majority of work that I do.

Justin Berman · « **Reply #15 on:** February 10, 2009, 07:47:14 am »

I seem to be in the minority here, but on my calibrated monitor. The posterization problem is all OVER the canon 5dmk2. I see problems in both files, and it honestly makes me wonder.

neil snape · « **Reply #16 on:** February 10, 2009, 08:24:16 am »

Quote from: Justin Berman

I seem to be in the minority here, but on my calibrated monitor. The posterization problem is all OVER the canon 5dmk2. I see problems in both files, and it honestly makes me wonder.

Ah if you can define the nature of the posterization it may help.
I have four calibrated monitors here , and don't really see much of a problem in posterization other than the fact the Canon still has a problem in dark red repro. The Hassy does not.
You must open the images in Photoshop or other clr mngd app as browsers aren't good at displaying images at 100%

Carsten W · « **Reply #17 on:** February 10, 2009, 08:35:09 am »

Quote from: neil snape

Ah if you can define the nature of the posterization it may help.
I have four calibrated monitors here , and don't really see much of a problem in posterization other than the fact the Canon still has a problem in dark red repro. The Hassy does not.
You must open the images in Photoshop or other clr mngd app as browsers aren't good at displaying images at 100%

The posterization that I was mentioning is also clearly visible whether in my browser (Safari, which does handle colour spaces) or CS4. I am talking about the facial shadow area on the left side of the image, especially around her right eye and the corner of her mouth. The shadows are so bad as to be useless here. My original 5D didn't give me results like this, and I don't understand why it is happening at all. I cannot imagine the results being accepted for critical use by any magazine? Is it always like this? The rest of the image looks fine, but that area is horrible.

lisa_r · « **Reply #18 on:** February 10, 2009, 09:32:01 am »

Carstenw, I see no issues in those shadows. (try a print and see what shows.) Clicking back and forth between the Canon and the H, the shadows look the same on my monitor...though the Canon files is a wee bit lighter. If you pulled those shadows down a hair to match the density of the H, they might match more precisely. I do see some unevenness in her skin tone which may be mistaken as posterization in BOTH files. (on her right cheek and her chin - left side of monitor.)

Carsten W · « **Reply #19 on:** February 10, 2009, 10:26:43 am »

Quote from: lisa_r

Carstenw, I see no issues in those shadows. (try a print and see what shows.) Clicking back and forth between the Canon and the H, the shadows look the same on my monitor...though the Canon files is a wee bit lighter. If you pulled those shadows down a hair to match the density of the H, they might match more precisely. I do see some unevenness in her skin tone which may be mistaken as posterization in BOTH files. (on her right cheek and her chin - left side of monitor.)

With "no issues", do you mean that it won't show in the final output, or that you don't see anything untoward even in these crops?

It looks to me like there is next to no information in those shadows, and the colour jumps very suddenly from black to dark red to normal skin tones.

Author Topic: H39MKII vs Canon 5DMKII image crop comparison (Read 17342 times)

lisa_r

lisa_r