[Brightcolours]

Exactly right, digital cameras (be it DSLRs or mirrorless somethings) are bigger (fatter) because film did not need anything behind it other than the pressure plate. So, being all nostalgic about how small some old SLR was (even though they were way worse to hold/operate) is quite useless.

[photonius]

Well, let's face it, there are a lot of reasons. If we look at the Canon SL1, they can be made small and compact, even though it's APS-C, but the mount is a full frame mount. Probably for little extra space/size, it could be made a FF camera. It's a marketing decision not to make such a small FF camera, because FF is more expensive, so it's being put into Pro bodies, with better sealing, extra slots, etc. etc.
If we take the old film bodies like EF-1, or A1, yes, they are flatter, because of the film thinness. Take the space on one side for the film role, and use it for a battery. Take the space of the other side film roll space, it's for memory chip, and maybe a processor. Overall, they can be made similar in size like the old days, give a few millimetres for the thickness of sensor and cooling. But due to modern material (plastics), they are lighter than old film bodies.
Why are they still so big? Many small reasons:

Screens on the back side, take that away, you save some space.
Articulated screen on the back side, it will be a bit thicker than a fixed screen (see change from Canon SL1 to Sl2).
Touch screen, yes, at's a tiny bit as well.
Space behind the thumb for all those wheels buttons - depending on what you wan to offer, you need a certain real estate. Note that the SL1/SL2 reduce the number. In film cameras you didn't have to worry about that.
Viewfinder: small conveniences: built-in flash (forget about that in old film cameras), eye sensors. Electronic overlays. Improved exposure meters. All needs to be squeezed in that bump on top. Ok, it's minimal, but nevertheless. For the point and shot cameras of the film days, they could shrink by taking away the mirror box, and the pentaprism, replacing it with a cheesy little window.
Handgrip: Yes, look at the cameras, all these huge hand-grips for convenience. Didn't exist on the EF or A1.
So, while bodies could be made smaller, user/marketing demands keeps them big.

As to lenses, yes, they are increasing as well. IQ demand is increasing. Compare the FD 50mm f1.4 versus the EF 50mm f1.4. Both have similar lens element design. The EF is a bit bigger, mainly in diameter, because the motors for AF had to be put in there as well. But the EF 50mm f.1.4 is still relatively compact. While many accept it's limitations, how many nowadays complain about it's problematic motor, it's border weakness wide open, etc. etc. So, look at the Sigma ART 50mm f1.4. Improved IQ, but you have to stick a lot of extra lens elements in there to achieve that. There is no free lunch. The only lenses where this doesn't apply are the tele lenses, there is enough space to put extra elements in there. Maybe they just get a bit heavier.
So, there is a trend for increased lens size because of quality demands. Users are less likely to accept a compromise only to get a small lens. Mirrorless, because of the reduced flange distance, can help for lenses in the 20-50mm range, but otherwise I think we are stuck with every increasing lenses for top quality lenses, or small lightweight kit stuff with limited apertures.

So, small can compact could be done, but would the resulting trade-offs (omissions) be acceptable to most people nowadays?

[Brightcolours]

If we look at the Canon SL1/2 we see that, even though it is APS-C, it still is fat behind the image plane. In fact, just as fat as a 6D.
http://j.mp/2IFwaM1
It can't be as small (slim) as film SLRs used to be (for above mentioned reasons).

[miro]

Thank you for all responses.  
I’m sorry for my late reaction. My initial message was not clear and only Claus Borisbg and partly Photonus tuch it in.
My approach is other way around. I don’t specify lenses, appretures,  formats and MPixesls. I talking about user needs.  My cuurent needs are
- Tack sharp A4 images that I print for amatur excepitions and sometimes place at home.
- Good IQ to display on cuurent and new 4K TVs
- Good IQ to show them at PC screen mobile Phone/ and social media
- Descent and fiture proof  video capability
- Nice to have – lighter and cheaper setup
What I use now. My equipment is 10..15 years old and it still fulfil most of my requirements – except video and lightweight.
Note I don’t need FF and 1,4 lenses, APS-C and 2,8 are more than enough. . I had fast lenses in the pass in bluur phobia time. Later on learned to make better composition and I would say that 100mm/2,8 is more than enough for my shoots. I never use my 150/2,8 @2,8. Put it simple no one likes my narrow DoF images except technical guys.
The only reason for FF is that I need PC/TS lenses and there is no such lenses for APS-C.

I was really wonder who really needs this FF bodies and razor-sharp 1,4 primes

[obican]

Some PC/TS lenses make even more sense on APS-C actually. You can stitch them and go for ridiculous resolution numbers, out of reach of any FF system. Sometimes I even prefer the x1.5 crop factor, unless I'm shooting architecture. In that case, Canon TS-E 17mm is the absolute king. Especially with the x1.5 Teleconverter when needed.

[JJ_SO]

I'm sorry, I've a hard time to follow this logic, obican. But maybe I think the wrong path along?

Each FF also can do APS-C size shots at roughly the same pixel density, meaning, can perform the same stitching - which on the other hand is something for static objects. Exclusively static objects... with preferably no light change during the whole shot collection for the Gigapixel panorama. Btw. which printer can do these?

Not to mention the contradictionary relation of camera weight an size plus tripod, nodal-point device and probably motorized tripod head, because moving the cameras in two or three horizontal rows is a rather dull thing to do. Not to mention PC time to collect and compute a nice stitch. With transformations needed, so interpolation comes into play as well and gone are the rectangular clean resolutions.

Also, I know a Canon shooter using a 100 Sonnar for his Gigapixel panos - in architecture. Now you try to tell me, using a teleconverter on a ultra wide is the thing to do? No metabone reducer, a teleconverter? Don't know about the Canon types, but Nikon and Fuji would crash the rear element of an ultra wide angle, such as a 17mm is.

Edit: Alright, on second thought I got the idea you're shifting the PC lens from right to left, so no nodal-ado. And so what? The PC lenses are made for FF, so I take a 19 or 24 mm and do the same trick to even more ridiculously high res numbers - but for what?

[Brightcolours]

You can't sensibly tilt and then just rotate the camera around to make huge stitched images, tilting the camera bites the purpose of tilting the lens. Same goes for shift...
And I wonder what you can do with APS-C what you can't do with FF or any kind of "medium format", concerning stitching? I must be overlooking an obvious angle that you are thinking of, obican.

[obican]

Let me go through it one by one.

1. You CAN shift the lens around, take individual shots and stitch them successfuly. You can't tilt them while doing that, that's true but shifting works just fine. And yes, static subjects only but you'd be surprised how well this actually works in many scenarios.

2. Yes the files are huge. If I remember correctly, a 24MP APS-C camera would yield a 120MP stitch.

3. If you use an adapter to keep the lens in place and shift the camera body around, instead of mounting the body and shift the lens around, the stitch will be pretty clean. No interpolation, no nodal change, no need for a motorized tripod head.

4. You are simply overlooking that most APS-C cameras have a much higher pixel density than FF cameras . It is true that the difference is not that dramatic anymore but there is still a difference. When you are shifting the lens around to "scan" the image field generated by the TS/PC lens, a higher pixel density will give you a higher resolution in the end. Right now every major company has an APS-C body which has a higher pixel density than any of their FF models. Again, used to be even more dramatic since you could've simply gotten a 24MP APS-C body but the best you could do with FF was 36 or even 24MP.

5. As for APS-C x1.5 crop and why I sometimes prefer it, simple. Sometimes I'd like to have a deeper DOF at a certain aperture and APS-C crop gives me that. It's about one stop more DOF at the same aperture with the same composition, since the camera has to be positioned further away. That's one stop less power on my speedlights. Of course I could just shoot them on a FF and crop later but then I lose the resolution. My A7 is merely 10MP when cropped to APS-C but even a cheap old Nex 7 could've given me 24MP with the same crop.

6. Teleconverters and tilt shift. This is tricky. First of all, forget everything I've said above. If you are shooting architecture and you need those ultra wide angles, you either go for a 17mm or a 24mm TS/PC lens. APS-C is simply no-no. So, you got your FF camera, you got your 17mm lens. You need the 24mm but don't have the money or don't want to carry it around? You can actually use Canon's x1.5 Teleconverter on their TS-E 17mm. Quality is actually quite close to what their excellent TS-E 24mm II can do, and you get even more shift range. It's a pretty neat trick, not sure if you can do the same with Nikon's PC lenses.

[Brightcolours]

You can do all that with FF, which will (depending on sensor resolution) result in more detail. There is a 17mm, a 24mm, a 45mm, a 50mm, a 90mm and a 135mm TS-E lens, so pretty easy to get more or less similar FOV whether you use FF or APS-C...

You can just stop down the aperture more and get the same DOF as APS-C too. And if you worry about "less power on your speedlights" then just use a different ISO setting?

[JJ_SO]

Let me go through it one by one.

1.  You CAN shift the lens around, take individual shots and stitch them successfuly. You can't tilt them while doing that, that's true but shifting works just fine. And yes, static subjects only but you'd be surprised how well this actually works in many scenarios.

2. Yes the files are huge. If I remember correctly, a 24MP APS-C camera would yield a 120MP stitch.

3. If you use an adapter to keep the lens in place and shift the camera body around, instead of mounting the body and shift the lens around, the stitch will be pretty clean. No interpolation, no nodal change, no need for a motorized tripod head.

4. You are simply overlooking that most APS-C cameras have a much higher pixel density than FF cameras . It is true that the difference is not that dramatic anymore but there is still a difference. When you are shifting the lens around to "scan" the image field generated by the TS/PC lens, a higher pixel density will give you a higher resolution in the end. Right now every major company has an APS-C body which has a higher pixel density than any of their FF models. Again, used to be even more dramatic since you could've simply gotten a 24MP APS-C body but the best you could do with FF was 36 or even 24MP.

5. As for APS-C x1.5 crop and why I sometimes prefer it, simple. Sometimes I'd like to have a deeper DOF at a certain aperture and APS-C crop gives me that. It's about one stop more DOF at the same aperture with the same composition, since the camera has to be positioned further away. That's one stop less power on my speedlights. Of course I could just shoot them on a FF and crop later but then I lose the resolution. My A7 is merely 10MP when cropped to APS-C but even a cheap old Nex 7 could've given me 24MP with the same crop.

6. Teleconverters and tilt shift. This is tricky. First of all, forget everything I've said above. If you are shooting architecture and you need those ultra wide angles, you either go for a 17mm or a 24mm TS/PC lens. APS-C is simply no-no. So, you got your FF camera, you got your 17mm lens. You need the 24mm but don't have the money or don't want to carry it around? You can actually use Canon's x1.5 Teleconverter on their TS-E 17mm. Quality is actually quite close to what their excellent TS-E 24mm II can do, and you get even more shift range. It's a pretty neat trick, not sure if you can do the same with Nikon's PC lenses.

First of all, thanks fot the interesting input to look a bit more into PC lenses. Especially the "keep the lens in Position and scan the scene by shifting" is a cool idea.

I think there are a couple of 42, 46, 50 MP FF bodies which would deliver a higher pixel count. So, by using the same distance of shifiting a FF HighRes body will get you more pixels - or less shft ops.

As for "not enough DoF": my current recipe to get the DoF I want is also not stitching but stacking.



Don't look to close, I didn't invest much time in cleaning the zones of Affinity's confusion  as railings, spokes or nets always will be a problem.