FULL FRAME or SUPER 35 - What do you prefer and why?

[jcs]

Great job Will! I see photos B&D (boy) form a nice stereoscopic 3D pair (shot side-by-side). If you 'look into the page staring at B & D you'll see a nice 3D image 

araucaria is pretty good at analyzing bokeh, etc.; guessed correctly in my two image tests- perhaps the (post) image reduction makes differences harder to see (and no tell-tale bokeh to analyze).

After studying medium & large format camera systems, the biggest difference in those camera designs vs. full frame and smaller is the image plane. In full frame & smaller, the lens is bigger than the image circle projection onto the sensor. This is why full frame can do better in low light: light is focused into a smaller area (and why the Speedbooster / focal reducers increase light performance by about 1 stop). Medium & large format cameras put the image plane much further back. The lenses aren't much bigger than full frame (if at all?) and appear to be much smaller than Super 35 cinema lenses. With the image plane so far back, these cameras+lens systems act like focal enlargers (apparently photocopier lenses can be used for photography?).

While many believe larger sensors do better in low light, that's not the case for medium and large format cameras with effective focal enlargers. They require vastly more light and/or very long exposure times. Found a post where someone noted they could see the bones in their hands when the flash went off. With their eyes closed! These larger format cameras also provide built in tilt shift capability, which helps deal with too much shallow DOF and/or reducing distortion (as with tilt shift lenses for full frame). The tilt shift is built into the camera, so all lenses will work.

It's clear that medium/large format lenses will have an advantage for shallow DOF for large objects/distances, and since tilt shift is built into the camera, are very useful for architecture and landscape. However, given that they need tons of light and/or long exposures, they're not as useful for moving objects.

In summary, it looks like medium & larger format provides more resolution potential (including microcontrast), easier shallow DOF (sometimes a problem for small objects up close), and tilt shift to help reduce distortion. I haven't found any examples which show the larger formats provides a unique look except, once again, more options for shallow DOF with larger objects. The same pattern comparing full frame to smaller sensors.

There's a lot of overlap between sensor size formats. Where the sensor size + available optics overlap, the images will be equivalent, where variances are due to lens designs. In the corner cases, larger sensors have advantages over smaller sensors with shallow DOF options for large objects (and also creates issues for smaller objects up close). Larger sensors, up to full frame, where the image circle is a reduction, can do better in low light. Once the sensor + optics acts like a focal enlarger, much more light is needed: medium & larger format. If there are other visible + measurable characteristics, it would be helpful to see and know what they are.

[SleepyWill]

Come on man the last two shots dont even have the slightest change in perspective, and for the other two are perfect color and character match. You are the perfect camera operator who can put cameras one a tripod without moving it at all and who can match two different cameras with different lenses to have exact the same color. Seriously you should go to hollywood they will pay you bigtime.

​I used a monopod, and you don't know who I am or what I do. You don't know who I work with or what I've worked on, nor how much I earn. Besides which, I don't think they are matched all that well actually. In fact the blacks are really bad, shamefully bad!

[Nick Hughes]

They're not same image- you can test yourself using photoshop and the "difference" blend mode. I overlaid A with every image in the third example. The first box is black because it's an exact match. None of the other images disappear because they are all slightly different. Certain points line up and cancel out (it's very close after all, that's the point of the test), but no matter how you move, resale, tilt the images, they never cancel out completely.

[jcs]

Come on man the last two shots dont even have the slightest change in perspective, and for the other two are perfect color and character match. You are the perfect camera operator who can put cameras one a tripod without moving it at all and who can match two different cameras with different lenses to have exact the same color. Seriously you should go to hollywood they will pay you bigtime.

​See the bricks in the periphery- there are differences. Nick's example makes the differences more clear.

[Nikkor]

Just cross your eyes as JCS said. Try to do the same at home, matching two different cameras with two different lenses, lol even with the same sensor size and focal length, you wont be able to pull it off.

[SleepyWill]

Just cross your eyes as JCS said. Try to do the same at home, matching two different cameras with two different lenses, lol even with the same sensor size and focal length, you wont be able to pull it off.

​Feel free to ask someone you trust, who has known ability to reproduce this test. I've just put in a great deal of effort to offer evidence that what I'm saying is accurate, but if even that's not enough for you, I guess you'd have to be with me at the time while I did it and even then, I bet you would find some other way to argue. I'm pretty much going to ignore you now, I hope you do the same for me. For everyone else, the challenge still stands! 

[Nikkor]

Shure man, I mean, monopod seriously? I love u too.

btw, you can upload the raw files if you want (not jpgs), that would rule out fakes.

[Cinegain]

Ok, listen, I know you can get a similar depth of field for the same equivalent focal lenghts although different sensor sizes. I'm not arguing that. But how about actual rendering of subjects? People love their full frame 85-135mm range for portraits... the interesting thing is... this choice is not a matter of depth of field, it's a matter of actually rendering shapes and giving off a flatering look. It's the projected lens image circle on that white piece of theoretical paper that was talked about earlier that's particularly being liked. If you take a 50mm lens you get different rendering on that piece of paper and using a smaller piece of paper isn't going to make your 50mm lens render like that one lens that has the focal length equivalent to the 50mm used on the crop sensored camera. So even if you match the depth of field, won't the rendering still be different, meaning you can't exactly replicate the full frame look all the time (which probably would become more noticeable the wider you then compare, being able to get real close otherwise)? So then it becomes a game of 'would you rather?'... would you rather... shoot a portrait (well, let's take a 1080p video with both cameras) with a M43 camera (2x crop) and a 45mm lens at f/2.8 or would you go with a full frame camera shooting at 90mm f/4 or would you claim it doesn't matter? Would you argue that the larger sensor surface doesn't give you richer results? How about lens performance? Don't you get a better optical performance out of your lens if you close it down a bit to f/4 instead of using it fairly open at f/2.8? I mean... you can throw in theoretical examples and mathematical equations... but real life deals with more parameters. Now... you can narrow down the error by having your subject far away and using a longer equivalent focal length to base your tests on... but get something a little bit closer and shoot a little wider... would you still say the same.

To get back to the posted 'proof'. I also have a little trouble with understanding how you used a monopod and four cameras to create these four images at the same time...

I mean, you could've just been taking a video with one camera only (with monopod, sure) and extracting 4 stills, each 1 frame apart (so much for the 'difference blend mode'). Because they're scarily similar in shooting position, detail rendering and everything. I mean. If you can match four cameras (or three cameras and a smartphone, is that what you were hinting at?) like that with same kind of dynamic range... contrast, colors, saturation and everything... then I tip my soft protective helmet to you, sir.

[jcs]

While we wait for example images showing larger formats provide unique looks beyond corner cases for DOF vs. their smaller brethren, here are some helpful charts & graphs from: http://www.falklumo.com/lumolabs/articles/equivalence/ff.html

 

Where quality is resolution, bokeh, artifacts, etc. The author is on the same page regarding equivalence and sensor size and no unique looks for sensor size. They created these charts & graphs to show how sensor size relates to what's currently available in lenses (at the time article was written).

[jcs]

Ok, listen, I know you can get a similar depth of field for the same equivalent focal lenghts although different sensor sizes. I'm not arguing that. But how about actualrendering of subjects? People love their full frame 85-135mm range for portraits... the interesting thing is... this choice is not a matter of depth of field, it's a matter of actually rendering shapes and giving off a flatering look.

All we need are example comparative shots where the cameras are set up for equivalence to show that a larger format renders the image somehow differently beyond what is predicted by the math (lens variance, etc.). 

[SleepyWill]

Because they're scarily similar in shooting position, detail rendering and everything. I mean. If you can match four cameras (or three cameras and a smartphone, is that what you were hinting at?) like that with same kind of dynamic range... contrast, colors, saturation and everything... then I tip my soft protective helmet to you, sir.

​Not a smartphone As for shooting position, my son was having an absence seizure (He has Doose syndrome, hence the soft protective hat), which was lucky because it saved you from having to see my ugly mug on the skintones "smooth" rendering test. Detail rendering, and everything else - I deliberately resized to obliterate methods other than sensor size characteristics to tell, especially as I know these cameras instantly from a 1 to 1 crop, I guarantee you guys do too. They are photographs, not video clips. When someone gets it right you can see the originals and you may be a lot less impressed. Do remember, if I was trying to cheat, I would have just shown ungraded ff and graded crop sensor, stuck subtle lights around the place stuff like that as all I am trying to demonstrate is that things like "smooth rendering" and "good separation" as claimed by monkey puzzle, when he finally conceded there is nothing unique about the ff fov or dof are a function of the lens and the sensor technology, not the sensor size.

[Cinegain]

You can go and count the number of question marks in my post. I'm a mirrorless crop guy myself (no experience with 35mm full frame here (my history with interchangeable lens cameras is a little something like this: GH2, E-M1 (soon to be gone), GH4, BMPCC and E-M5II (soon to be added)... do have something DX, the D5300 and the image is great, thing is... when used to mirrorless cameras (especially liveview (features) and EVF) working with the D5300 feels like a bit of a step back)). Was wondering a bunch of stuff and playing a bit of devil's advocate in the process. Just the other day someone told me that 45mm on M43 for portraits is not a great idea (he's been into cameras and lenses for decades), so that got me thinking... why is that? Is it the rendering? As we're discussing the same here, I threw it in the mix.

That said. Depth of field aside, to me there's just something about full frame results I've seen that I like and think is pretty magical. Maybe it's the 5D's softness that has something familiar and yesteryears about it, reminding of old school film, with 4K and rich detail crispness being the future perhaps we are still adjusting to (kinda the same reason people are split on HFR?). Would've jumped aboard the S.S. 5DmkIII if it wasn't for the same reasons I'm not really loving the D5300 (which has nothing to do with the image quality/results you can get with it)... so basicly just that mirrorless is much nicer to work with from the perspective of having more user-friendly features and what is super important to me: you can go crazy compact... no lugging around huge and somewhat heavy gear (although it's not so much the weight as the size that restricts me)... less obtrusive... but at the same time being able to meet the requirements in more demanding situations, easily adapt into whatever set-up you need it to be, allowing for mounting just about any lens you can think of as well. Not really that on board with the whole 'full frame is cheaper' either. Go start out with a bit of money. Maybe you could just manage to get the 5DmkIII with 24-105mm f/4 and then that's it. Go ahead and buy some more (native mount) full frame glass. That don't come cheap, yo. It's another reason I started out with a cheaper GH2 and mostly added cheap but great vintage lenses from eBay to my collection (or just cheap native electronic ones). Not really seeing how going full frame is the cheaper route here. But I'm drifting off...

[jcs]

Cinegain- I started out on this journey with the Canon 5D Mark II. I gave my dad my first DLSR, a Canon 40D (which only shot stills). For years I thought there was something special about full frame. It wasn't until the GH4 came out did I learn about equivalence, otherwise I wouldn't have purchased an m43 camera. Being able to shoot shallow DOF with the Voigtlander 25mm F.95 and GH4 in a relatively small package is very cool. I still felt there was something magical about the 5D (Mark III at that point) and full frame beyond shallow DOF. After using the A7S and comparing it to the 5D III, I realized it was Canon's color science that was providing some magic (RAW is reasonably sharp). For video we want the most detail possible without aliasing. We can always add softening filters if needed (I use a Tiffen Black Pro Mist 1/4 on the A7S, which is sharper than 5D 3 ML RAW). Doing the S35 to full frame equivalence challenge removed all doubt: sensor size alone does not provide a unique look. The full frame look is a... better described as the shallow DOF look. I feel for the folks who don't believe it yet. It took a while to research and do the tests to reach this conclusion.

[leeys]

Uh, guys. Of course you can match DoF, perspective and FoV. Each format's own advantages and disadvantages comes down to size and image quality. The latter is more noticeable in stills, I feel. If you haven't seen what you can do with an 8x10 contact print, you should try it out sometime.