Andrew Reid

Let's stick to doing the math horizontally only this time as it matches better what our own eyes see when we switch between 16:9 full frame 1080p and 16:9 cropped 4K video on cameras like the 5D Mark IV (1.74x crop for UHD).

The X-T2 and GH4 have extra crop factors in 4K due to their sensor readout... Say the X-T2 extra crop is a 1.17x crop of a 1.5x sensor What is the correct maths here? Is it - 1.5 * 1.17 = 1.75 Or 1.5 + 0.17 = 1.67 ?

No the GX85 does not use the full 4.6K sensor resolution, it crops a 4K portion of that The entire sensor is still a 2x crop compared to full frame, just as it will be on the GH5. The DVX200 is a news gathering ENG style camera with documentary and faux-cinema leanings, but the main thing is that those who use it want to record for longer and need small file sizes. That's probably why it doesn't have 10bit 4:2:2 at high bitrates. It isn't about the GH5 getting a better codec than a more expensive camera... It is about matching the codec to the type of user. GH4 users have spoken very loudly about wanting 10bit 4:2:2 internally and not just on the HDMI. If Panasonic were serious about segmenting their cameras based on 8bit and 10bit, they probably wouldn't have put a 10bit output on the GH4 in the first place!

This is going into an article so it has to be right... As others have pointed out, when a camera does a 1:1 readout of the sensor in 4K mode, you can JUST use the horizontal resolution to calculate the added 4K crop of the full frame 16:9 window of the sensor. This way the math is really easy - just by dividing the total horizontal resolution of the sensor by the resolution of 4K can you get the crop factor. 4096 for DCI 4K or 3840 for UHD So I tried it on a few cameras and it gives this - Canon 1D C (DCI 4K) - 1.26x (5184 / 4096) Canon 1D X Mark II (DCI 4K) - 1.33x (5472 / 4096) Canon 1D X Mark II (UHD) - 1.42x (5472 / 3840) Nikon D5 (UHD) - 1.45x Panasonic GH4 (DCI 4K) - 2.12x Panasonic GH4 (UHD) - 2.2x How is this mathematically, does it stack up? What is interesting is how little difference there is between the 1D C and 1D X Mk II in DCI 4K after all... I think most people were quoting the UHD crop when it first came out. The field of view difference here we are not talking the difference between 3:2 full frame and cropped 4K. We are talking the difference between 16:9 full frame 1080p and the cropped 1:1 4K window.

Ah yes specs are so meaningless Policar.... After all that's why you shoot on an Alexa!!

Metabones say "native contrast detect AF" but I detect no improvement on the A7S II, it's still unusable with my Canon glass for any AF work in stills mode and video racking forget about it However very curious to see if their native mode works well on the A7R II! It works well on my A6300... Very well indeed!

I have an A7S II and sold my A7R II. It was a close call though. Now I know recently the A7R II had a big firmware update (v3) mentioning phase detect AF in movie mode and other improvements. A7S II is only V2 and it doesn't have phase detect AF on the sensor like A7R II. A7R II was already very fast in stills mode even with some Canon lenses. Now also the Metabones adapter has had a big firmware update adding native AF support - which should mean a huge increase in AF performance with Canon EF lenses on Sony E-mount cameras and extra features like Eye AF, DMF and better movie mode AF. Has anyone tried both firmware updates together, with their Canon lenses on the A7R II... Does it amount to a big improvement and more solid AF reliability? Usable in movie mode too? With the 5D Mark IV's only real selling point being the native EF mount and Dual Pixel AF, the better the situation gets on the Sony bodies, the happier I will be!

If the rumours turn out not to be true there's no reason to be too unhappy... Panasonic could for example bridge the gap in quality between the 8bit GH4 and the 8bit 1D C before going all out with a fancy 10bit codec. Better low light performance and a better less noisy LOG profile is a hope of mine A simplified selection of picture profiles each with a separate use would be good, like this - - A vivid punchy one with high colour saturation - A more standard muted one for light grading - A LOG profile - A flat profile for those who want to adjust the gamma curve in post without going all out with LOG and LUTs Just 4 or 5 would do it. Get rid of the fluff! Also let's see them go back to the wide aspect ratio 1.86x crop sensor, like the GH2, great for DCI 4K 4096 x 2160 1080p 120fps with no crop and no moire would be FANTASTIC All this is within the bounds of current technology, they just have to get it RIGHT!

Not enough times for some people, judging by the way they are STILL going on about this again and again. Here's how Philip Bloom has started responding to this "just use a cinema EOS / insert video camera name here" bullshit - "thats such a boring, dull, oft repeated, lazy. apologist and close minded comment! Smaller cameras like this are incredibly important...gimbal work and documentary work for example PLUS the whole joy of 5Dmk2 when it came out was FULL FRAME. There are NO full frame video cameras. I should have this reply on copy and paste." Ever since I started EOSHD people have been saying "just use a camcorder" blah blah blah it's a stills camera not a video camera It totally misses the point!

Just say "fast" shutter speed This 0 degree stuff is nonsense If we went on naming things in technically incorrect ways we'd be calling low sensitivity ISO 0 at this rate!

0 degree shutter? That doesn't make sense. That shutter would be closed. It does look like it is shot at a high shutter speed rather than an ND used... But hey it's only one of their main pre-release promotional videos so such details don't matter, obviously

No but low light performance might take a hit due to the 1:1 crop. Oversampling improves low light performance - look at the Super 35mm mode on the A7R II.

Good points those! Good point. YAGH was a really bad decision. It should always have been in the form of a battery grip

Sony have a 42MP sensor and at least they gave us a choice - Full frame sensor readout using pixel binning for 4K without crop Super 35mm crop from 5K sensor area With Canon they have a 30MP sensor and they are doing in 4K mode - A 1.74 crop only And a lack of Canon LOG is silly. This is a camera with a 5 year shelf life and it leaves them seriously exposed to competitors in that time frame to come... All that the others need to sort out is their own Dual Pixel AF as well implemented and as user friendly as on the 5D Mark IV.

The X-T2 1080p looks cleaner with less moire and aliasing than the X Pro 2. The 4K looks to have a bit of moire and aliasing going on but only if you pixel peep it on a chart. In real shooting situations rarely will it be an issue. The 4K from the X-T2 looks similar to the A7S II in terms of detail / moire / resolving power but softer than the A6300. I wonder if it holds up at high ISOs as well as the A6300 If it had the IBIS of the GX85, did not crop slightly in 4K and made the F-LOG profile available for internal recording it would almost be PERFECT. As it is... well just compare it to the X-T1... a massive leap up considering it is Fuji we're talking about!!

The Sigma 18-35mm is a far better option. F1.8 vs F4! Problem is in UHD at 1.74x crop you're going to get 32mm at the wide end rather than 28mm.

"Full frame" is a photographic standard so it measures 3:2 in terms of aspect ratio. So surely it is more correct to say "4K on the 5D Mark IV is a 1.74x crop of full frame" "Full frame" is not a 16:9 box. OK we can measure only horizontally and that is 1.64x crop But what about UHD 3840px? That would change the crop again and most people deliver in 16:9 So effectively the 5D Mark IV crop factor is more often than not 1.74x. Anyway no matter what I think.... I hope we can we get a clearer consensus on how the crop factor calculations break down - Reading the thread so far and various contributor math - DCI 4K measured diagonally from the 3:2 photographic area it is 1.74x crop DCI 4K measured diagonally from the 16:9 video area of full frame is 1.64x crop DCI 4K (4096 pixels wide) is a horizontal crop factor of 1.64x So what is the UHD 4K crop if your delivery needs to be in the usual aspect ratio and not the 17:9 of DCI 4K? Back around 1.74x measured horizontally? And what is the industry standard way of measuring video crop factors - the ones stated by manufacturers themselves - is it horizontally or diagonally? And do they take into account the full sensor 3:2 or 4:3 aspect ratio if measuring diagonally or just the 16:9 portion? And as for the lenses - EF-S mount lenses WILL NOT work on the 5D Mark IV as the rear glass intrudes too far into the mount... Although I hear on some lenses like the 10-22mm you can pop off the plastic shield at the back and it might work - no mirror hit? And so Tokina 11-16mm F2.8 is best wide for 4K and you would need to swap lenses when you go back into FF stills mode

I did skip maths class, it's fucking uninteresting. I am more interested in images But it is 1.74x crop from the full 3:2 sensor as Micro Four Thirds is a 2x crop from the same, so the 1.64 I can't accept, as it is measured from a different area to all the other cameras on the market

No but to keep things consistent with other cameras sporting different aspect ratios, it's the total sensor crop that matters not the crop from a letterbox 16:9 of the sensor For instance Micro Four Thirds is a 4:3 2x crop of a 3:2 full frame sensor Therefore to make the comparison worthy, 5D Mk IV 4K video is a 1.74x crop of the same Unless we stop talking about Micro Four Thirds 16:9 video as a 2x crop? But I feel that horse has somewhat bolted!

I think the problem is in LOG mode the Sony cameras apply a wide gamut to colour which leaves it at the mercy of compression and bad colour paths in post. The white balance...something is going on there too... If they had just adjusted the gamma curve like Canon LOG and left colour alone we wouldn't have all these threads and messed up grades in the first place so the blame has to lie at Sony's door really

If it were measured only horizontally then you could have a weirdly extreme letterbox aspect ratio and call it 1.0x crop if it measured the full width of the sensor, so that can't be how they calculate it surely? Full sensor area of the 5D Mark IV is 3:2

I am guessing the main difference are the teams behind them Looks like Canon have distinct stills and video teams who work separately on programming the chips These guys are primarily controlled by their bosses though And I am pretty sure their bosses follow a strict overall company strategy Which is to divide EOS photography lines and Cinema EOS lines very carefully so they don't cannibalise one another in terms of pro stills or video features

I hope the GH5 comes in 2016, it may be delayed to 2017. Indeed Super 35mm CMOS from Panasonic would be great. They have already made a full frame sensor for Leica so have the raw material in-house to do it Micro Four Thirds should be the APS-C of Panasonic's range Where would Canon be without full frame?!

Didn't Samsung put their Super 35mm CMOS from the NX1 on the open market? Why did Nikon or Canon not buy this, back it up with more Samsung chips as is common in the wider semiconductors industry and build their own camera around it? There are no excuses for Canon and Nikon not to push things to the limit.... I am tired of the lame excuses for products and their wide margins at the expense of innovation Frankly it is embarrassing. How much more evolution can you do with a 1970's SLR?

Hey Luke. enjoyed your posts but I believe the crop is actually measured diagonally not just horizontally? Vertically it is a 2x crop... 4480/2160 = 2.07x So average the vertical and horizontal and we meet in the middle... actually the exact maths I can't be too precise about but I guess it is an average of 1.64x and 2.07x? How do you measure the diagonal crop? 18-35mm would be a solution to get 31.32mm to 60mm... Not a very appealing zoom range To get a 24-70mm is a lot harder