To crop, or not to crop, 1.6x on the 5D!

Nick_C wrote in post #4470437
If you are talking about the 5D crop after resizing

No, before it was resized, the edges of the black outline seem to have some USM halos and if I run a USM pass on the 30D crop it then looks remarkably similar.

Are the raw files from both available to download somewhere?

pxl8 wrote in post #4470479
No, before it was resized, the edges of the black outline seem to have some USM halos and if I run a USM pass on the 30D crop it then looks remarkably similar.

Are the raw files from both available to download somewhere?

Yes you can apply USM to the image & it will no doubt sharpen up to the 5D image fairly well, but I didnt want to do that for the test as I wanted unprocessed images, both converted from raw files in exactly the same manner, nothing more, nothing less!

The whole point of this test was to see if there was any major differences or losses when cropping to the 1.6x factor & enlarging so the 5D crop ended up at 8mp, the same as the 30D, if I had applied USM to just one photo what would be the point of such a test as this? to make the 5D look better perhaps? what would be the point in that? im not quite sure what your getting at here

I'm just making an observation based on the images posted. The 30D image looks somewhat fluffy compared to those out of my own 30D and 40D whereas the 5D seems to be have been processed in some way. I've never seen an image come out of a raw converter with that kind of artifact before which I why I asked if they could be downloaded from somewhere.

With regards to the test would it not be a more realistic exercise to process both files to their greatest potential and then compare results? I would certainly question the merit of judging the printed 30D file if it hadn't been processed against the 5D file where you clearly made the effort to get the best result possible from the upscaling. Or to put in a somewhat simplistic manner, here's one I didn't bother with and here's one I did - which is best?

pxl8 wrote in post #4471206
I'm just making an observation based on the images posted. The 30D image looks somewhat fluffy compared to those out of my own 30D and 40D whereas the 5D seems to be have been processed in some way. I've never seen an image come out of a raw converter with that kind of artifact before which I why I asked if they could be downloaded from somewhere.

With regards to the test would it not be a more realistic exercise to process both files to their greatest potential and then compare results? I would certainly question the merit of judging the printed 30D file if it hadn't been processed against the 5D file where you clearly made the effort to get the best result possible from the upscaling. Or to put in a somewhat simplistic manner, here's one I didn't bother with and here's one I did - which is best?

Well as I said, these are unprocessed, if you think the 5D has been processed then thats your view I guess (I cant do anything about people not trusting the validity of the tests).

As for processing both to their best, well thats kind of like testing 2 lenses & processing the images to their best, sure you could make the £200 consumer grade lens get close to the £600 L lens, but you not seeing the TRUE image, its down to how well its been processed, which would have made this a completely pointless test.

The original crops are as I said unprocessed, the sharpening on the 5D "Upsized" image is no doubt performed somewhere in the program used to enlarge it, but I wanted to start with 2 images just processed in a very basic way from raw, no tweaks of any kind, it just happens that the 5D images look better to start with than those from the APS-C body, yes I could have sharpened up the image a little to match it to the 5D but that would be unfair as I should then have done the same to the 5D image, even then there would still be that sharpness gain by applying the same USM to both images as the 5D is already sharper which I have put down to the fact that lenses appear to perform better on the 5D than on the cropped bodies.

To better explain the halos I'm seeing I downloaded the first 5D raw file from this page:

http://raw.fotosite.pl …D_24-70L_f2.8_ISOtest.htm

I processed two versions of the file in DPP, one with the raw sharpening at 3 (default for the file) and again with it at 0. In the sharpened version there are clear halos around the black text that aren't present in the unsharpened version - see attached image.

These halos look very much like the ones I saw in your 100% crop which is what lead to the question in the first place. I've attached a crop and marked some of the halos.

What I don't understand is why the halos have appeared in your unprocessed file as a raw file without sharpening will be soft due to the AA filter built into the sensor.

What you could do about people not trusting validity of the tests is to make the test files available for people to repeat it for themselves and maybe even find a method that improves upon the work you've already done...

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

pxl8 wrote in post #4471667
To better explain the halos I'm seeing I downloaded the first 5D raw file from this page:

http://raw.fotosite.pl …D_24-70L_f2.8_ISOtest.htm

I processed two versions of the file in DPP, one with the raw sharpening at 3 (default for the file) and again with it at 0. In the sharpened version there are clear halos around the black text that aren't present in the unsharpened version - see attached image.

These halos look very much like the ones I saw in your 100% crop which is what lead to the question in the first place. I've attached a crop and marked some of the halos.

What I don't understand is why the halos have appeared in your unprocessed file as a raw file without sharpening will be soft due to the AA filter built into the sensor.

What you could do about people not trusting validity of the tests is to make the test files available for people to repeat it for themselves and maybe even find a method that improves upon the work you've already done...

This proves what I was saying, lenses perform better on the 5D than say the 30D

The test consisted of the subject, all the usual tripod, MLU etc.. best of approx 10 images taken on each body, using the same lens, resetting AF each time, not that there was much (if any) AF variation to be seen.

With the 24-105L on the 5D, images are punchy & contrasty & need no USM whatsoever, I dont use it anymore on ANY of my images, if I do they look too sharp, the same lens tried on a 350D & a 30D shows that the image is softer & needs more processing, its a similar story with any lens I have tried so far.

I guess to make it a closer comparison I should make up for the softness on 30D shots by applying USM, but this is just another point that I was trying to get across in this little test, why would I want to use say a 70-300IS on the 30D when the images are worse than from a 5D using the exact same lens? I have to carry another body around with me & all for what? the 1.6x gain? well I can easily achieve this with results that are just not worth me lugging another body around with me.

Yes I could host the Raw files somewhere, but I assumed trust wasnt an issue

Ever thought of a career in politics?

If a raw file from the 5D produces those sharpening artifacts with no sharpening applied in processing I for one want to see it.

What app did you use to process the files btw?

I just cant see where you are going with all this?

If its that you dont trust the tests fine, I dont mind, but all I can do is put up the results for all to see, I am not going to start processing images as you said to get the best out of each, not in a test like this, I will only ever shoot under controlled conditions & process both equally alike, if one comes out looking like its been sharpened, well so be it.. but zooming in to say 400% on a jpeg isnt a good way to scrutinize images.

The app was CS3.

Where I'm going is simple. You're saying the 5D is capable of producing images that need no capture sharpening whatsoever, the raw files I've seen don't seem to agree so I'd like to see one that confirms it as it would convince me to seriously consider purchasing a 5D.

pxl8 wrote in post #4471950
Where I'm going is simple. You're saying the 5D is capable of producing images that need no capture sharpening whatsoever, the raw files I've seen don't seem to agree so I'd like to see one that confirms it as it would convince me to seriously consider purchasing a 5D.

Perhaps you might find this of interest then,

http://www.kenrockwell​.com/tech/full-frame-advantage.htm

I dont much about this guy but if you hover over the samples you will see the difference im talking about, & yes you can even make out some halo's on the window frames

Not really, it's not from a raw file processed with sharpening at zero, it's a from an in camera jpg with the sharpening set to 3....

pxl8 wrote in post #4472072
Not really, it's not from a raw file processed with sharpening at zero, it's a from an in camera jpg with the sharpening set to 3....

Well then I cant help on this thread, the main purpose of this was about resizing, it was all about gaining the crop factor if you own a full frame body not a 5D vs 30/40D comparison

Wait until I get the website up & running, its 95% done, I will have plenty more tests on there you may find of interest, I will be sure to include some of my images that require no USM.

Great, I look forward to seeing it. I'm keen to see some raw files that really show what the 5D can do!

ok I plan to do a whole article on the benefits that I have personally seen over the APS-C bodies as I find this whole subject facinating