jcs

I think what people fail to realise here is that for most people using MF lenses (or any vintage lens) I'm sure has nothing to do with whether you can get other "equivalent" lenses to match or do an "equivalent" job... it's all about the things that aren't supposed to be there, the "mistakes" if you will that make the image interesting. It's like playing music, jazz as an example, the notes you don't play and the unexpected notes you do play make what you are doing interesting to some people. If you just played a super predictable structured melody I don't know many musicians personally who would be overly impressed...
That being said not everyone may enjoy what you are doing (they never will in life) but if something has worth and merit to you don't let anyone tell you otherwise.

I will give you the big secret, everything you do in your day to day life is pointless, in fact life is pointless. There's no plan waiting for you at the end, no one there to pat you on the back and tell you did life right. Just do what makes you happy and help others where you can  .

I had one with a set of mamiya on a a7s, sold it quickly.

First the focal reducer add a fair amount of barrel distortion,
Second, as stated by others, focal reducer.... reduce focal so your 80mm 2.8 become a 56 mm f2.... It s not so exciting, you can t go super wide either .
The only good point for me was that the lenses where bigger with a good focus throw, so quite production friendly.
Due to the modern market full frame lenses got a lot of improvement compared to medium format, today you can get pin sharp lenses wide open at 1.4, so getting medium format glass with a speedbooster on full frame is not such a good idea IMO. 

Using it without can be decent as you only use the center of the image circle but you will be mostly stuck at F2.8.
Keep in mind that the alexa 65 sensor size is not far off full frame either.

If you want the medium format feel on smaller format, just use lenses a bit longer and faster than you are use to, something like the Nikon 105 f1.4 

Ahem... Earspeakers  (for those who've never listened to them, click the link, they are amazing tech!). Stax + Carver Magnetic Field Power Amplifier* + organic matter = wow.
I've got a pair of old HD 580's which are super comfy though the Audio Technica ATH-M50's are more detailed (for mixing and critical listening). Sony 7506s are great all-around, smallish, light and comfy too.
 
* https://en.wikipedia.org/wiki/Bob_Carver Looks like he's now retired- amazing sound engineer!

It's a good theory, everything is a good theory.

Each of us runs a local simulation of reality- there can be a point if you want, or not. We are all the same

Well perspective compression is relative to your distance to your subject ( that why people speak about equivalence , a 42.5 on micro four third and a 85mm on full frame will produce the same compression at the same distance from the subject) 

The test you show simply say that the Leica are more corrected in term of distortion than the Cooke.

If you like distortion, get the Kippon, you will get a lot of it. But in term of compression and dof your 80mm f 2.8 will act like a 52mm f 2 .

The point of larger format is I guess that wide angle are a bit easier to manufacture.

I used to use an i1 Display calibrator on Windows and OSX, unfortunately for whatever reason the calibrations seemed to change (don't think it was normal hardware changing over time, seemed to be software related). I now use two factory calibrated 4K Dell monitors, a UP3216Q (main), and a P2715Q (secondary). When the image/video looks good on these displays it looks good on MBP Retinas, iPhones, iPads, Sony TV, Samsung TV, Apple Cinema display, etc. Still have the i1 Display calibrator- haven't needed it since upgrading monitors (probably wouldn't hurt to use it however given the fiddly stability results (could be the OSes), probably not worth the time. Maybe when people start looking like this
I'll put it back into service  ).

The challenge with *nix is there are so many flavors (versions).  Ubuntu and relatives are perhaps the mostly popular right now... What we do as software developers is download the source code for tools/apps and build for the specific version of *nix that we are using (this too can be a pain and time burner). This is great for back-end server products but not front end user-facing products. This would not be possible for mass-market commercial software like Adobe. The other challenge is UI development tools on Linux are currently very weak. Just debugging code on Linux with an IDE is currently very very weak. XCode on OSX and Visual Studio on Windows are still light years ahead of anything on Linux (Microsoft has released Code for Linux, but mostly for webapp development (not C/C++ which is needed for native apps)).
Working on Linux feels like working on a Model T vs. Porsche or Ferrari on OSX and Windows respectively (Windows still has the best developer tools, by far). C++ Builder used to run on Linux, however support was dropped: https://www.embarcadero.com/products/cbuilder . This is an incredible cross-platform toolset for C++ development. They could bring it back to Linux if there was a market. The challenge with Linux is most desktop Linux users don't pay for anything, and want everything for free (and/or pirate software). Until or if major software vendors can work out a model that works for Linux users (can't be totally free), don't expect any major products to be ported to Linux. AutoDesk and Blackmagic do support Linux, however that's for very vertical market applications (renderfarms are a good use for Linux, and having many seats in a (e.g. one-off movie) production environment where the OS is free can make financial sense).
I switch between OSX and Windows 10 frequently. Instead of dual-booting my 'old' 2010 12-core + GTX 980ti anymore, I built a brand new 10-core box (6950X) + GTX 1080 which I was going to make into a Hackintosh. After getting it running with Windows 10, I was so impressed with the superior font rendering, snappier GUI response, and ability to display a 4K desktop vs. OSX, as well as superior 4K editing performance and less crashes (really an NVidia driver issue- it is what it is), I dropped the idea of putting OSX on it. I use a special USB switch to switch keyboard and mouse input, and use the 2 4K Dell display's multiple inputs to switch between OSX and Windows 10 without having to reboot. Works really well in production.
I still use MacBook Pros running OSX, however modern software such as vMIX (for live streaming with efficient GPU acceleration) only run on Windows (not really enough space on the internal SSD to run Bootcamp at this point. It is possible to install Bootcamp on an external drive with a bit of effort (easiest with Thunderbolt, a bit more work with USB3), see YouTube for some examples).

So if we focus on the art, the emotion of the combined 'lens filter', we can categorize looks so others can replicate themselves* and quit arguing about sensor size. I think Mattias' pics look cool, and if a MF->FF focal reducer plus cheap MF lenses can be used to recreate those looks, that's pretty neat- gives new love to neglected unused MF lenses
 
* this was not meant as a cloning comment  

The challenge with *nix is there are so many flavors (versions).  Ubuntu and relatives are perhaps the mostly popular right now... What we do as software developers is download the source code for tools/apps and build for the specific version of *nix that we are using (this too can be a pain and time burner). This is great for back-end server products but not front end user-facing products. This would not be possible for mass-market commercial software like Adobe. The other challenge is UI development tools on Linux are currently very weak. Just debugging code on Linux with an IDE is currently very very weak. XCode on OSX and Visual Studio on Windows are still light years ahead of anything on Linux (Microsoft has released Code for Linux, but mostly for webapp development (not C/C++ which is needed for native apps)).
Working on Linux feels like working on a Model T vs. Porsche or Ferrari on OSX and Windows respectively (Windows still has the best developer tools, by far). C++ Builder used to run on Linux, however support was dropped: https://www.embarcadero.com/products/cbuilder . This is an incredible cross-platform toolset for C++ development. They could bring it back to Linux if there was a market. The challenge with Linux is most desktop Linux users don't pay for anything, and want everything for free (and/or pirate software). Until or if major software vendors can work out a model that works for Linux users (can't be totally free), don't expect any major products to be ported to Linux. AutoDesk and Blackmagic do support Linux, however that's for very vertical market applications (renderfarms are a good use for Linux, and having many seats in a (e.g. one-off movie) production environment where the OS is free can make financial sense).
I switch between OSX and Windows 10 frequently. Instead of dual-booting my 'old' 2010 12-core + GTX 980ti anymore, I built a brand new 10-core box (6950X) + GTX 1080 which I was going to make into a Hackintosh. After getting it running with Windows 10, I was so impressed with the superior font rendering, snappier GUI response, and ability to display a 4K desktop vs. OSX, as well as superior 4K editing performance and less crashes (really an NVidia driver issue- it is what it is), I dropped the idea of putting OSX on it. I use a special USB switch to switch keyboard and mouse input, and use the 2 4K Dell display's multiple inputs to switch between OSX and Windows 10 without having to reboot. Works really well in production.
I still use MacBook Pros running OSX, however modern software such as vMIX (for live streaming with efficient GPU acceleration) only run on Windows (not really enough space on the internal SSD to run Bootcamp at this point. It is possible to install Bootcamp on an external drive with a bit of effort (easiest with Thunderbolt, a bit more work with USB3), see YouTube for some examples).

So if we focus on the art, the emotion of the combined 'lens filter', we can categorize looks so others can replicate themselves* and quit arguing about sensor size. I think Mattias' pics look cool, and if a MF->FF focal reducer plus cheap MF lenses can be used to recreate those looks, that's pretty neat- gives new love to neglected unused MF lenses
 
* this was not meant as a cloning comment  

We're on the same page.

This discussion reminds me of similar ones for analog vs. digital audio, vinyl/tape vs. digital, tubes vs. MOSFETs, one preamp vs. another preamp, one mic vs. another, crisp and clinical vs. warm and creamy, etc. Head on over to GearSlutz for a taste of audio drama for those inclined: https://www.gearslutz.com/board/ (around 10 million posts!). It's a great resource for audio related questions for filmmaking too.
In agreement with what @Brian Caldwell just said- these lens combos are creating interesting character, analog optical filters (transfer functions), which is art, and that is cool. It has nothing to do with sensor size per se, only the combination of optics to get the desired (or discovered!) look.
Here's an example of Shane Hurlbut comparing Leica Summicron-C to Cooke S4, same sensor size (Super35), and getting vastly different results (his perception, some of you may even disagree with his results):
http://www.thehurlblog.com/lens-tests-leica-summicron-c-vs-cooke-s4-film-education/
To wrap it all up:
Leica Summicron-C Lenses:
Flatter image Makes Monette look more sophisticated and older Makes her look wider than the Cooke S4 More of a white out with lens flares; doesn’t do as well as the Cooke S4 with lens flares. Bokeh less stop signing, more round Neutral lens Less detail in her face Doesn’t hold highlights as well as Cooke S4 Less contrast
  Cooke S4 Lenses:
More three dimensional quality Makes Monette look younger Makes her look skinnier Better with lens flares than the Summicron-C lenses Slightly more yellow than the Summicron-C Bokeh has more of a stop signing effect More detail in her face Holds highlights more than the Summicron lenses More contrast Thanks @bunk for creating these excellent computer renderings, which physically simulate perfect lenses and sensors via ray tracing photon paths for different sensor sizes and using perfect mathematical equivalence with pixel perfect results:
http://www.eoshd.com/comments/topic/20975-full-frame-aesthetic/?do=findComment&comment=167154


@Mattias Burling offered $100 if the images matched perfectly (if I read the thread correctly, see link above). Mattias did you pay bunk? The time and effort to set up and render those examples was worth way more than $100, and should end any further arguments regarding sensor size for anyone interested in a scientific and rational discussion (bunk also included the Cinema 4D project files for anyone to replicate as per the Scientific Method). @gatopardo replicated the results with VRay (another photon ray tracer) and 3DSMax.
Mattias' comments on medium format lenses being very cheap and 'free' in some cases is very useful information for those admiring the kinds of looks possible with a focal reducer and old medium format lenses, which create interesting analog filters which a lot of people love. As noted by Hurlbut and others, the Cooke S4 causes distortion and has other artifacts which make people look thinner, more 3D, better highlights, etc. Imperfection is the spice of art- makes the flavor interesting which will be loved by many where different forms of spice will be loved by others (or for some uses, no spice at all- clinically accurate (Sony/Zeiss for example)). Like comparing Neve/Avalon preamps (warm/colored) to Grace (clinically clean). Neither better than the other, chocolate vs. vanilla...

hahaha true.

This discussion reminds me of similar ones for analog vs. digital audio, vinyl/tape vs. digital, tubes vs. MOSFETs, one preamp vs. another preamp, one mic vs. another, crisp and clinical vs. warm and creamy, etc. Head on over to GearSlutz for a taste of audio drama for those inclined: https://www.gearslutz.com/board/ (around 10 million posts!). It's a great resource for audio related questions for filmmaking too.
In agreement with what @Brian Caldwell just said- these lens combos are creating interesting character, analog optical filters (transfer functions), which is art, and that is cool. It has nothing to do with sensor size per se, only the combination of optics to get the desired (or discovered!) look.
Here's an example of Shane Hurlbut comparing Leica Summicron-C to Cooke S4, same sensor size (Super35), and getting vastly different results (his perception, some of you may even disagree with his results):
http://www.thehurlblog.com/lens-tests-leica-summicron-c-vs-cooke-s4-film-education/
To wrap it all up:
Leica Summicron-C Lenses:
Flatter image Makes Monette look more sophisticated and older Makes her look wider than the Cooke S4 More of a white out with lens flares; doesn’t do as well as the Cooke S4 with lens flares. Bokeh less stop signing, more round Neutral lens Less detail in her face Doesn’t hold highlights as well as Cooke S4 Less contrast
  Cooke S4 Lenses:
More three dimensional quality Makes Monette look younger Makes her look skinnier Better with lens flares than the Summicron-C lenses Slightly more yellow than the Summicron-C Bokeh has more of a stop signing effect More detail in her face Holds highlights more than the Summicron lenses More contrast Thanks @bunk for creating these excellent computer renderings, which physically simulate perfect lenses and sensors via ray tracing photon paths for different sensor sizes and using perfect mathematical equivalence with pixel perfect results:
http://www.eoshd.com/comments/topic/20975-full-frame-aesthetic/?do=findComment&comment=167154


@Mattias Burling offered $100 if the images matched perfectly (if I read the thread correctly, see link above). Mattias did you pay bunk? The time and effort to set up and render those examples was worth way more than $100, and should end any further arguments regarding sensor size for anyone interested in a scientific and rational discussion (bunk also included the Cinema 4D project files for anyone to replicate as per the Scientific Method). @gatopardo replicated the results with VRay (another photon ray tracer) and 3DSMax.
Mattias' comments on medium format lenses being very cheap and 'free' in some cases is very useful information for those admiring the kinds of looks possible with a focal reducer and old medium format lenses, which create interesting analog filters which a lot of people love. As noted by Hurlbut and others, the Cooke S4 causes distortion and has other artifacts which make people look thinner, more 3D, better highlights, etc. Imperfection is the spice of art- makes the flavor interesting which will be loved by many where different forms of spice will be loved by others (or for some uses, no spice at all- clinically accurate (Sony/Zeiss for example)). Like comparing Neve/Avalon preamps (warm/colored) to Grace (clinically clean). Neither better than the other, chocolate vs. vanilla...

Character/rendering are entirely separate issues from photographic basics like DOF and perspective.  Certainly, if you like the character then there's nothing wrong with that.  Heck, if you find that using an 80mm lens with both a 1.4x teleconverter and 0.7x focal reducer (in series!)  to get back to 80mm but with some funky aberrations gives you the results you want then you should certainly do it.  My only real objection is with pseudo-explanations - basically "fake physics" - that cross the line into mysticism.
Mini correction:  techically, aberrations can and do influence DOF.

I remember Northrup being ridiculed here on this forum. JCS was the  one who came to his “defense” and made the article he linked to in his post one page back. It’s simple math, it’s how things work.
It’s not a coincident that Houdini, Max, Maya, SoftImage, Modo, Lightwave, Blender, you name them all will give the same outcome when you set up a camera test. They all use the same math from the beginning of 3D time.
Hollywood uses it to produces blockbusters. The test I made was completely clean but you can add distortion to mimic specific lenses and maybe even sensors (wouldn’t surprise me).
I’m not a lens specialist but I don’t think a 25mm F0.7 exists in the real world, yet I used it in the example above. But that is not the issue here.
The question was what are the full frame aesthetics compared to the others. And they answer should be the technical limitations we have to deal with in every day live. But those will be solved and when that happens, the answer is none.
Then the believers will start to point out the differences in characteristics of lenses and or sensors but like Gatopardo said they are there between FF lenses and sensors as well …and since there will be more computer generated lenses within time than vintage lenses, those differences will fade as they will all give the same dull characteristics and you won't be able to tell what camera or lens was used ...and I'm talking about consumer cameras as this is a consumer forum.

Same findings here in VrayCam in 3DsMax. In real world images would always be somewhat different because we can't use the same lens in both bodies. So, the differences will be on the same level as the ones comparing a Canon 50mm 1.4 to the Nikon 50mm 1.4 one. It's each lens distortions and character making the difference rather than equivalence. But this is important because we may not be able to find a specific character of a full frame lens on a "equivalent" crop sensor lens.

Thanks for creating the animated gif which clearly shows equivalence, even when the settings aren't exact (which mostly effects shadows- bokeh matches pretty nicely). If you want to see perfection, dig up the computer graphics renderings from one of the "FF/MF look" threads on this forum: the results are pixel perfect (probably (photon) ray-traced optics).
I think what people are liking about the MF lenses is their character, and even what some might call artifacts (including artifacts from the focal reducer).

Of course results are what count. My 2 cents: just don't get caught up in a possibly expensive quest for medium format glass assuming there is a super different, unique, mythical look to be achieved. (Not to say that hunting down and using vintage glass is a blast!).
Understanding equivalence can help one understand what results can be expected from any combination of sensor size and lens. Yes, there are unique combo's at the extremes of the range (no m43 equivalent lens for a 50mm f1.0 on full frame). But in general: for the most frequently used focal lengths, you can find equivalent lenses for the current different sensor sizes, thus yielding (close to) identical optical results.

Thanks for creating the animated gif which clearly shows equivalence, even when the settings aren't exact (which mostly effects shadows- bokeh matches pretty nicely). If you want to see perfection, dig up the computer graphics renderings from one of the "FF/MF look" threads on this forum: the results are pixel perfect (probably (photon) ray-traced optics).
I think what people are liking about the MF lenses is their character, and even what some might call artifacts (including artifacts from the focal reducer).

You are expecting a level of precision in this comparison that is entirely unreasonable.  Little things like changes in distortion and entrance pupil position during zooming make it impractical to make a blink comparator test completely perfect.  What the comparison does show - with more than sufficient precision - is that you can optically reproduce all aspects of an image shot on a large format with one shot on a smaller format - or vice versa.  
The notion that, say, an 80mm medium format lens has some inherent "80mm-ness" or "medium formatishness" that somehow stays with that lens after you attach a focal reducer is just silliness.  The combination of a 0.7x focal reducer and an 80mm lens is a 56mm lens.  Period.  Put that 56mm lens on a 24x36mm format camera and it will behave just like any other 56mm lens attached to that camera, the only caveats being related to aberrations and other flaws in the lens and focal reducer.

Thanks for creating the animated gif which clearly shows equivalence, even when the settings aren't exact (which mostly effects shadows- bokeh matches pretty nicely). If you want to see perfection, dig up the computer graphics renderings from one of the "FF/MF look" threads on this forum: the results are pixel perfect (probably (photon) ray-traced optics).
I think what people are liking about the MF lenses is their character, and even what some might call artifacts (including artifacts from the focal reducer).

You should be aware that this is a religious discussion.  Comparisons, and discussions about comparisons, should be avoided and only discussed in private!

I did those tests- it's not really possible to do perfect equivalence with physical lenses unless all the settings can be exactly matched. In the first example, the only major difference was shadow detail which could be related to ISO... In the second example, maybe I made a mistake or it's still related to optics not really being equivalent. The 'normal' test matches almost perfectly.
Brain Caldwell, the optical engineer and inventor of the Speed Booster says the same thing regarding FF vs. MF. That's why he wasn't interested in making a MF to FF SpeedBooster...
In any case, the differences are minor and most people couldn't tell the difference. Someone posted computer graphics (ray traced?) examples that matched perfectly, as the math predicted.