Jay60p

I agree. I would add that "constant quality" (variable bitrate) works best for me.
Here’s what has surprised me over the past two years with mirrorless video:
I shoot 4K 60p in-camera at 200Mbps H265 10bit.  When I re-compress this to H265 10bit In Handbrake (at many times slower than real time) at the RF 19 setting with “constant quality”, I get near identical image quality at only 10Mbps to 60Mbps depending on the subject matter.
The scenes at higher bit rates (50-60Mbps) are outdoor shots with a lot of complicated textures from trees, grass, etc. But indoors shots with solid color walls and less busy backgrounds can come out as low as 10-20Mbps.
I compared test files like this:
Open the test movies in Quicktime player windows.
In one of them go to the menu “Window / merge all windows”. 
This puts them all in one window with tabs along the top for each video.  
Find the same frame In all the movies, and set all at the same magnification.
Do an instant back & forth comparison by clicking on the tabs.
Comparisons at high magnification shows a tiny bit of softening in the Handbrake image compared
to the original, not visible at normal magnification, at a fraction of the of the original camera bitrate.
It seems to me that the speed of compression processing affects the IQ as much as the bitrate.
In other words, letting Handbrake run a compression slowly at 5 frames a second (resulting in a 20Mbps variable bitrate “constant quality” file) can give the same IQ as the original camera file did in real time at 60 frames per second (at 200Mbps “constant bitrate”.)
More time/frame makes sense when you look at what happens in the algorithm settings Handbrake uses for compression as you change the slider under encoder options from fast to slow.
For example, h264 10bit, slider at “medium” shows the following settings:
x264 Unparse: level=4.0:vbv-bufsize=75000:vbv-maxrate=60000
compared to slider at “very slow”:
x264 Unparse: level=4.0:ref=5:bframes=8:b-adapt=2:direct=auto:analyse=all:me=umh:merange=24:subme=10:trellis=2:vbv-bufsize=75000:vbv-maxrate=60000:rc-lookahead=60
Going from 200Mbps to 20 or 30Mbps for comparable IQ was a surprise.
This is probably common knowledge for Handbrake users, but I haven’t seen it discussed (or I missed it).
Happily this means I can archive an excellent master at a fraction of the original 200Mbps.
And fit everything on a flash drive for the TV’s USB slot.
To get back on topic, I usually go at least five years between cameras, which means I might still have
three more to go with the X-T3. I have oversampled 4K60 DCI 10bit, Fuji fixed the auto ISO stepping, I got the lenses and accessories I need so I’m fixed indefinitely.
Now if Sony 4K projectors would come down in price…

I agree. I would add that "constant quality" (variable bitrate) works best for me.
Here’s what has surprised me over the past two years with mirrorless video:
I shoot 4K 60p in-camera at 200Mbps H265 10bit.  When I re-compress this to H265 10bit In Handbrake (at many times slower than real time) at the RF 19 setting with “constant quality”, I get near identical image quality at only 10Mbps to 60Mbps depending on the subject matter.
The scenes at higher bit rates (50-60Mbps) are outdoor shots with a lot of complicated textures from trees, grass, etc. But indoors shots with solid color walls and less busy backgrounds can come out as low as 10-20Mbps.
I compared test files like this:
Open the test movies in Quicktime player windows.
In one of them go to the menu “Window / merge all windows”. 
This puts them all in one window with tabs along the top for each video.  
Find the same frame In all the movies, and set all at the same magnification.
Do an instant back & forth comparison by clicking on the tabs.
Comparisons at high magnification shows a tiny bit of softening in the Handbrake image compared
to the original, not visible at normal magnification, at a fraction of the of the original camera bitrate.
It seems to me that the speed of compression processing affects the IQ as much as the bitrate.
In other words, letting Handbrake run a compression slowly at 5 frames a second (resulting in a 20Mbps variable bitrate “constant quality” file) can give the same IQ as the original camera file did in real time at 60 frames per second (at 200Mbps “constant bitrate”.)
More time/frame makes sense when you look at what happens in the algorithm settings Handbrake uses for compression as you change the slider under encoder options from fast to slow.
For example, h264 10bit, slider at “medium” shows the following settings:
x264 Unparse: level=4.0:vbv-bufsize=75000:vbv-maxrate=60000
compared to slider at “very slow”:
x264 Unparse: level=4.0:ref=5:bframes=8:b-adapt=2:direct=auto:analyse=all:me=umh:merange=24:subme=10:trellis=2:vbv-bufsize=75000:vbv-maxrate=60000:rc-lookahead=60
Going from 200Mbps to 20 or 30Mbps for comparable IQ was a surprise.
This is probably common knowledge for Handbrake users, but I haven’t seen it discussed (or I missed it).
Happily this means I can archive an excellent master at a fraction of the original 200Mbps.
And fit everything on a flash drive for the TV’s USB slot.
To get back on topic, I usually go at least five years between cameras, which means I might still have
three more to go with the X-T3. I have oversampled 4K60 DCI 10bit, Fuji fixed the auto ISO stepping, I got the lenses and accessories I need so I’m fixed indefinitely.
Now if Sony 4K projectors would come down in price…

And here is the best basis for selecting a computer, the operating system you are happiest with.  Most important I think.
Mac users have traditionally paid more for Apple hardware than PC users paid for Windows hardware, precisely for the same reason, that they prefer the OS. (I am one of those.)
But now for the first time (?) Mac users can pay a little less.
 

And here is the best basis for selecting a computer, the operating system you are happiest with.  Most important I think.
Mac users have traditionally paid more for Apple hardware than PC users paid for Windows hardware, precisely for the same reason, that they prefer the OS. (I am one of those.)
But now for the first time (?) Mac users can pay a little less.
 

Recently most of the forum topics have been about new FF cameras for $3-5 thousand, but I bet 90% of views on this forum are from people looking to spend $1000 or less.
Even in the USA the average family savings account is only about the price of a new car, and for age under 35 the average is $10,000.
If you do mostly hand held shooting and don't care about 4k60p, the X-S10 is the obvious choice for the IBIS.
 
I compared the After Effects stabilizer (same as Premier?) doing walking/running torture tests, and the Final Cut Pro InertiaCam stabilizer was much better.
I haven't tried Resolve.
 

Recently most of the forum topics have been about new FF cameras for $3-5 thousand, but I bet 90% of views on this forum are from people looking to spend $1000 or less.
Even in the USA the average family savings account is only about the price of a new car, and for age under 35 the average is $10,000.
If you do mostly hand held shooting and don't care about 4k60p, the X-S10 is the obvious choice for the IBIS.
 
I compared the After Effects stabilizer (same as Premier?) doing walking/running torture tests, and the Final Cut Pro InertiaCam stabilizer was much better.
I haven't tried Resolve.
 

according to this interview
(https://www.imaging-resource.com/news/2020/09/04/fujifilm-interview-covid-gfx-strategy-shrinking-ibis-300K-cycle-shutter)
X-H2 will come with "some sort of breakthrough", "something more revolutionary".
Possibly global shutter? oversampled 6K60? AI boosted AF?
Since they appear to be very ambitious for the X-H2, I am willing to wait. X-T3 is still all I need now.

Yes, thank you, but where is it?!
They have to act fast, most people in business I know will get a Canon from now, till Christmas. Canon will "dominate" the video market again.
It has been at least 2 years since the XH1, it wasn't a huge success, I thought they learned by that release, "smaller" companies have to do a couple of right moves to put them ahead the curve, BM does alright, Fuji had a few nice releases lately, but they have to be pro active, and get ahead.
Also, they need some better "video" lenses, less breathing, some kind of parfocalness, I.S on hybrid lenses (a better 16-70mm 4f, this one is a very expensive dissapointment), the excellent 10-24 on a newer version, e.t.c
Their ecosystem is 95% photocentric right now, and do not tell me about their 2 manual MK lenses, they are nice, but not enough, and certainly not up to what we expect in 2020-2021.

A few thoughts on this topic:
1) I would have expected this equivalency theory would have been tested more reliably by still photographers at the numerous
photography forums long ago. They use a much wider range of format sizes than the video people here at EOSHD.
If not, it could be there is just too many variables to control, or no consensus on the methods to use.
 
2)  I would suggest using a 4x5 sheet film camera (8x10 is at $15 a shot!) and limit the test to manual lenses.
Mount all lenses on a 4x5 lens board and take a 4x5 shot for each, to be scanned for viewing.
This way the camera does not change, the sensor does not change, no digital transformations are done in camera.
The different lenses would have different size image circles in the 4x5s, so would be of different resolutions,
but that should not effect the depth of field comparisons much.
 
I did look at the SLR primes with the Turbo II speedbooster. It shrinks the first fringing seen, but it includes more of the edges of the
image circle, with more CA, so overall the fringing looked the same. It really is not a big problem, in video it will never be noticed unless
you look for it, it's more of a problem in still photos.  I use these SLR primes for stop-motion and time lapse, where you don't want any
communication with the camera that changes the lens settings.
 
Here is a review of my favorite Fuji lens that includes comments on the in-camera corrections (CA, vignetting, distortion) for anyone
unfamiliar with this: https://opticallimits.com/fuji_x/887-fuji1024f4ois?start=1
 
My question is, what about the third party lenses? Do the mirrorless cameras generally apply in-camera corrections to
Sigma, Tokina, Tamron, that come in their lens mounts?
 
 
 
 

Yes, it is a very good thing in the Fuji lenses. That's why I was asking about third party lenses, if they don't get corrections that puts them at a disadvantage.
And I wonder where a particular lens's correction information is stored, in the lens firmware or the camera firmware? Just curious.

There is another factor with the new digital cameras which will complicate DOF/format test results.
My Fuji X-T3 recognizes specific Fuji lenses and automatically corrects aberrations for those lenses in camera.
I have FF Canon, Takumar, Minolta 35mm prime lenses with dumb adapters, and none of them are as sharp and clean as the two APS-C Fuji zoom lenses I use. The Fuji zooms don't show color fringing. My FF 35mm primes and Nikon F zooms do (on very close examination).
 

Yeah, I put off installing new firmware for a month or two, sometimes it breaks more than it fixes.
Good to hear a happy ending!

To test this you could use:
standard 16mm, Bolex (Kern Paillard 10mm at f/1.8
4/3rds, 17mm at f/3.2
APS-C (Nikon DX), 22mm at f/4
Full Frame, 34mm at f/6.3
8x10, 256mm at f/45
The 16mm would have to be wide open and the 8x10 would have to be completely stopped down (my 8x10 270mm is f/4.4 - 45)
This is according to this calculator:
https://www.pointsinfocus.com/tools/depth-of-field-and-equivalent-lens-calculator/#{"c":[{"f":13,"av":"8","fl":50,"d":3048,"cm":"0"}],"m":0}
Damn it you’ve got me doing it now!
Like Oliver Hardy and his taxi driver getting involved in Stan Laurel’s jigsaw puzzle,
and they miss his wedding ceremony…

I think the biggest drawback for larger sensors is they have always been a few years behind the smaller sensors in performance, at least in mirrorless hybrids.
My most important spec is 4k60p. Fuji had it in 10 bit HEVC two years ago.  The FF Sony & Canon 10bit took two years to catch up & match it. And they are still nowhere near the $1500 the X-T3 was. Panasonic's GH5 4K60p also looks great from two years ago, I still have MP4 samples I downloaded late 2018 that look indistinguishable from FF 4k60p.
It used to be in the old film days that a doubled increase in format size gave a doubled increase in resolution. So when I first heard DSLRs were about to go full frame I thought, Great!  But, in digital it's not that simple. Video UHD is still 3840 x 2160 no matter the sensor size.
The other major drawback is price - this from Wikipedia...
"Production costs for a full-frame sensor can exceed twenty times the costs for an APS-C sensor. Only 20 full-frame sensors will fit on an 8-inch (200 mm) silicon wafer, and yield is comparatively low because the sensor's large area makes it very vulnerable to contaminants—20 evenly distributed defects could theoretically ruin an entire wafer."

To test this you could use:
standard 16mm, Bolex (Kern Paillard 10mm at f/1.8
4/3rds, 17mm at f/3.2
APS-C (Nikon DX), 22mm at f/4
Full Frame, 34mm at f/6.3
8x10, 256mm at f/45
The 16mm would have to be wide open and the 8x10 would have to be completely stopped down (my 8x10 270mm is f/4.4 - 45)
This is according to this calculator:
https://www.pointsinfocus.com/tools/depth-of-field-and-equivalent-lens-calculator/#{"c":[{"f":13,"av":"8","fl":50,"d":3048,"cm":"0"}],"m":0}
Damn it you’ve got me doing it now!
Like Oliver Hardy and his taxi driver getting involved in Stan Laurel’s jigsaw puzzle,
and they miss his wedding ceremony…

I don't recall having that happen, tried changing some things but those settings haven't greyed out.
Is this a new camera?

I think the biggest drawback for larger sensors is they have always been a few years behind the smaller sensors in performance, at least in mirrorless hybrids.
My most important spec is 4k60p. Fuji had it in 10 bit HEVC two years ago.  The FF Sony & Canon 10bit took two years to catch up & match it. And they are still nowhere near the $1500 the X-T3 was. Panasonic's GH5 4K60p also looks great from two years ago, I still have MP4 samples I downloaded late 2018 that look indistinguishable from FF 4k60p.
It used to be in the old film days that a doubled increase in format size gave a doubled increase in resolution. So when I first heard DSLRs were about to go full frame I thought, Great!  But, in digital it's not that simple. Video UHD is still 3840 x 2160 no matter the sensor size.
The other major drawback is price - this from Wikipedia...
"Production costs for a full-frame sensor can exceed twenty times the costs for an APS-C sensor. Only 20 full-frame sensors will fit on an 8-inch (200 mm) silicon wafer, and yield is comparatively low because the sensor's large area makes it very vulnerable to contaminants—20 evenly distributed defects could theoretically ruin an entire wafer."

One large format, coming right up... 8x10 is only available in anamorphic cinemascope sir, there's no demand for any others.

 

One large format, coming right up... 8x10 is only available in anamorphic cinemascope sir, there's no demand for any others.

 

Unfortunately the 16-55 does do stepping during a zoom. It is the lens being used in the first video I linked above at time 4:52.
That is how the lens maintains a constant aperture, same as my Fuji 10-24 f4 (see my post above, Thursday August 6).
I was thinking the 18-55 was doing some sort of ISO stepping, but putting the lens to my ear I can hear faint clicking that
sounds like aperture stepping while zooming at above f4. So the new X-T4 firmware may not fix this.
The OIS is strong in this lens, great for hand holding in one position. It may be overly strong for panning shots since there is little
ease-out when coming to a stop and appears too abrupt. So I pan slowly with OIS on.
I have several old 35mm SLR zoom lenses that work best for zoom shots. They have manual aperture rings (no stepping) and are very parfocal so they don't rely on the camera's auto focus.  They are also a small fraction of the cost of new Fuji zooms, and use $20 adapters.

One other condition avoids zoom exposure stepping:
If you have the X-T3 settings all on manual and your 18-55 is wide open at f2.8 (at wide angle), then a zoom
does not do the exposure stepping and will darken at the telephoto end during a video shot.
But close the aperture down to f4 or more and the stepping will happen 3 times during a zoom;
and the telephoto will end at the same brightness that the wide angle started at.
Note that 3 steps x 1/3rd stop each = one stop = from f2.8 (wide) to f4 (telephoto).
For still photos this is an advantage, for video it is an unwanted "blinking" artifact.
 
 
 

I started with mirrorless video a year and a half ago and I'm still learning how it works.
Recent discussions about new cameras and image quality got me wondering about
how much oversampling cleans up noise in high ISO shots. Especially in Super35.
So I shot this test on my X-T3, ISO 25600, 18-55 zoom at f8 on tripod from the same position. Varied shutter for exposure. Samples straight from camera without any grading. (First time I ever shot at this high ISO!) Natural room light from windows with Sunlight white balance.
First - actual size crop from a “FINE” JPG photo at 6240 x 3512 for a 1:1 (not oversampled) comparison. Eterna profile with boosted color, noise reduction -4 and sharpness -4, lowest.

 
Second - actual size crop from an oversampled UHD 60p 10bit HEVC movie frame, same settings:
 

 
the 6K is from this angle:
 
and the 4K oversampled is from this angle:
which required a little zoom lens trial and error to get the "actual size" crops equal to compare.
 
I also recall Andriodlad said something about the Fuji 60p having a little less oversampling than 24p or 30p, they would be cleaner.
 
 
 
 

Yes, with Bayer sensors, 1:1 readout is always softer than oversampled readout. Strong OLPF that combats moire can make it even softer.

Yes, family video is 90% of what I do nowadays,  and 4k 60p is the best format to use to capture the most live You-Are-There video.
Another thing I did is replace all the lamps in my house with 5000K Led bulbs. That solved major white balance problems caused
by window light mixing with indoor lamps, and solved some major color grading headaches.  They looked strangely cool for a few days,
but we became used to them fast (the old lamps look horribly yellow now).
On X-T3 I use the Eterna film simulation at DR 200, but boost the color level in camera. I find it needs no further grading if exposed
correctly at the right white balance, a major time saver.
 
 

Yes, family video is 90% of what I do nowadays,  and 4k 60p is the best format to use to capture the most live You-Are-There video.
Another thing I did is replace all the lamps in my house with 5000K Led bulbs. That solved major white balance problems caused
by window light mixing with indoor lamps, and solved some major color grading headaches.  They looked strangely cool for a few days,
but we became used to them fast (the old lamps look horribly yellow now).
On X-T3 I use the Eterna film simulation at DR 200, but boost the color level in camera. I find it needs no further grading if exposed
correctly at the right white balance, a major time saver.