The 100% Crop Thing

[Toby]

I, as many others, have read all the threads regarding the quality of the M8 sensor. Most often the 100% crop has been presented as a measure of empirical quality. I, amongst others have expressed doubts you can see anything from such a test, but I'd not really given any thought to why this should be so. I think what I neglected to consider that this is not a like for like test against cameras. All a 100% crop shows is the quality of a file at maximum enlargement at 300 DPI or 72 DPI or whatever. In other words if you look a 100% crop of a RD1 file with a 100% crop of an M8 file they won't look very different but if they were prints the RD1 print would be a 8x10 whilst the M8 print would be closer to 11x14! In other words the test is not comparable across cameras and so proves nothing. If anything the 100% crop is only really indicative of lens resolution with different lenses on the same camera. I'm by no means a tech head so if this is wrong please feel free to (politely) correct me but a comparable test of the would be against a 125% enlargement of an 8mp camera or a 75% enlargement of a 12mp camera at the same resolution (dpi). What are anyone elses thoughts?

 

[Nachkebia]

Just don`t forget to take in considiration viewing distance....

 

[markinlondon]

Thanks for asking this, Toby, I've been wondering about this for a while. And Nachkebia has hit it too, viewing distance is crucial in may ways. Perhaps we should be referring to radius of enlargement as I think that's what we're really talking about. As a film dinosaur, when testing film I'll usually take a sectional enlargement to the equivalent of about 16x12 (12r) on 10x8 paper and force myself to view it from at least a meter to assess grain and tonality. The "100% crop" thing seems a bit too much like the old camera club "shove your nose in the print" assessment for my liking.

 

[Toby]

Thanks for asking this, Toby, I've been wondering about this for a while. And Nachkebia has hit it too, viewing distance is crucial in may ways. Perhaps we should be referring to radius of enlargement as I think that's what we're really talking about. As a film dinosaur, when testing film I'll usually take a sectional enlargement to the equivalent of about 16x12 (12r) on 10x8 paper and force myself to view it from at least a meter to assess grain and tonality. The "100% crop" thing seems a bit too much like the old camera club "shove your nose in the print" assessment for my liking.

Yes, viewing distance is critical but I'm trying to think of a meaningful test that could be viewed next to each other online.

 

[markinlondon]

I see your point, Toby but as usual there are too many variables to my mind. Digital cameras and/or scanners do so much processing of the image without your intervention that you can never be sure that no "transparent" tinkering has happened to produce the image you see on screen. Add to that the massive choice of sensor sizes and recording options and I find it a minefield. At least with a 35mm negative you're starting from the same size original.

 

[Trius]

Why not just set the image size the same (as well as all the other workflow steps) for all your tests, regardless of the camera/sensor?

All this pixel-peeping is funny, anyway, IMO. Years ago I attended an Ansel Adams exhibit. Like a fool, I had waited till the last day of the show so was there with a large crowd. That meant as I moved through the exhibit I was more or less trapped with any morons that happened to be next to me.

So this youngish guy (prolly in his late 20s or so) was there with his dad, and I couldn't help but overhear the kids inane comments. When we were all in front of a large (30x40 or so) print from the 1940s, t'he kid got right up to the print and said to his dad, "Yup, that's 35mm."

At that point, I couldn't stand it any more, and said "That is NOT 35mm." The kids said "You don't think so?", and I told him that I KNEW so, then proceeded to tell both of them why.

Not slagging anyone who is trying to evaluate image quality, and I realize $5K is a big chunk of chain for a camera unless you're a high volume pro and it's a business expense, but sheesh, can't we all just enjoy photos?

 

[jaapv]

Yes and no- basically a 100% crop is the maximum enlargement one will get without pixellation on the monitor. So screen size is a factor as well. In general- for photographs- I agree it is rather meaningless, but when one wants to make some technical point - like I wanted to show a M8 file at maximal resolution with the maximum of information visible-,
I do think a 100% crop makes sense. After all, one just cuts a 800x600 piece out of the picture and displays that.

 

[Toby]

Yes and no- basically a 100% crop is the maximum enlargement one will get without pixellation on the monitor. So screen size is a factor as well. In general- for photographs- I agree it is rather meaningless, but when one wants to make some technical point - like I wanted to show a M8 file at maximal resolution with the maximum of information visible-,
I do think a 100% crop makes sense. After all, one just cuts a 800x600 piece out of the picture and displays that.

But an 800x600 crop is not the same percentage of the image across different sensors its like taking a two 35mm prints one 8x10 the other 16x20 taking a 2 inch square from both of them and buying a camera based on those crops.

 

[markinlondon]

Actually Toby I think it's more like cutting a square out of a print from (say) a Leica neg and a Minox neg that have been enlarged to give the same size print. You'd buy the Leica on that test as you'd buy the 10x8 view camera if you compared that to the Leica (I'm assuming we're working on a "I want the one that did that" basis here). I still think there's too much going on between capture and screen to make an objective judgement.

 

[jaapv]

But an 800x600 crop is not the same percentage of the image across different sensors its like taking a two 35mm prints one 8x10 the other 16x20 taking a 2 inch square from both of them and buying a camera based on those crops.

Correct. I'm not advocating comparisons on that basis though, it is about displaying (part of ) the picture at maximum resolution, which can make sense to look at certain aspects. Just look at it this way: if you had a 4000x3000 pixels screen, you would not need 100% crop.

 

[Toby]

Correct. I'm not advocating comparisons on that basis though, it is about displaying (part of ) the picture at maximum resolution, which can make sense to look at certain aspects. Just look at it this way: if you had a 4000x3000 pixels screen, you would not need 100% crop.

I think I'm going to live in a cave and make cyanotypes

 

[markinlondon]

I think I'm going to live in a cave and make cyanotypes

You wouldn't get enough sunlight in a cave for cyanotypes 🙂

 

[jaapv]

I think I'm going to live in a cave and make cyanotypes

`

Most caves are taken by trolls already🙁 You wouldn't fit in the scene.....

 

[Benjamin Marks]

FWIW, I think we are entering a time in which the qualitative differences among sensors and in-camera sharpening software are diminishing. So 100% crops may be useful to illustrate _a_ difference between cameras, but not one that is going to have a huge effect in, say, a 6x9 print. Obviously there are a lot of variables here to consider. I am getting very usable images from my RD-1; I also get very nice (and substantially larger) files from my 5D. I am sure that the top-of-the-line Canon would be another bump in quality. But because differences at my most common print sizes are small, a 100% crop on screen may be one way to illustrate the resolving power of one system compared to another. Useful? Who knows.

 

[JohnL]

All this is true. 100% crops have their uses but figuring out how a print is going to look at any given enlargement is not really one of them. You can get some pointers, but that's about it. If, OTOH, you have the entire files, you can print them both at whatever size you like and probably draw some meaningful conclusions.

 

[rbiemer]

This is why I think the "100% crop" isn't very useful for any thing,
I've attached a "100% crop" of one of my latest photos. What, you don't see anything? Oh, right, sorry. I cropped 100% of the photo so there's nothing left...😀
Rob

 

[Trius]

Oh. I thought it was just a lenscap shot.

 

[Quinn Porter]

Yes, viewing distance is critical but I'm trying to think of a meaningful test that could be viewed next to each other online.

Ultimately, several different tests are needed to compare various cameras of different pixel counts and sensor sizes. Generally, it is useful to compose the same image in each camera, then downsize the larger file to match the pixel count of the smaller one, and finally, compare crops. The downsizing has the effect of reducing noise and gives you a better indication of how the cameras would compare if prints were made at the same size.