Does APS-C Crop Mode Reduce Video Quality? (Sony A7 IV)

Not the most incredible production quality, but I put a little effort in. Hopefully this helps some people out there who have been misinformed.

[@HairSuitGentleman]

Hey man, your presenting skills are great and your video is very well structured! You should make more!

[@Niberspace]

Wow dude, I can't believe KZbin finally has a video guy who actually understands cameras. It just makes me so tired to see KZbinrs saying the same horrible advice. The number of KZbinrs I've seen who keeps advising people to use crop-mode almost makes me wanna quit videography, they keep using arguments like "It both outputs the same resolution anyway" and "as long as you're not in a tight room where you can't step back then there's no drawback to using crop-mode"

[@TechnoBabble]

@@Niberspace yeah, I hate that whole "you still get 4k". That doesn't mean it resolves the same amount of actual detail, plus the extra noise gives you worse dynamic range

[@stevenstallings324]

Thanks for the informative video! No apologies necessary - your presenting skills and the quality you achieved in this video are excellent! Currently, I am debating between the A74 and the A7R5. very different cameras to be sure; however, they'd be used for the same work - video/photo hybrid use (mostly shooting 23.9 fps) with 32 MP being the minimum req for my photo work. After selling my a7siii, I'm left with a Ninja V, and of the two current options I'm giving myself, the 7R5 has the capability of shooting RAW, which I sometimes do on certain, very specific, productions. Anyway, this video helped me in my little consumerism detour of researching cameras, and I thank and virtually applaud you for the work you've done, good sir. Also, I'm applying your findings to the new A6700, and do wonder how the APSC crop on these two FF cameras differs from it, as opposed to the A6400. Cheers!

[@TechnoBabble]

Thank you very much!

[@VincentDelToral]

very useful tests!

[@arhon888]

Thanks for sharing. This is a very useful comparison. It is worth noting, that when comparing HD vs HD crop on the A7IV, the quality and dynamic range actually improves. The same was true of my Canon 90d.

[@TechnoBabble]

Yep, that's because for whatever reason Sony decided that the camera wasn't going to downscale the full sensor readout for full frame 1080p but it WAS going to downscale the 4.6k readout of the APS-C crop for it's 1080p. Even more odd is that this is only the case up to 60fps, at 100 and 120fps the full frame 1080p is actually better quality than the APS-C crop.

[@robertconrad6802]

fun fact the 4k crop still has more detail than a7siii ff 4k, but certainly a bit noiser. [a1, a7rv included as well]. a lot of the current down sampling tech in the industry is really great for bird/nature videography given you have enough light to stay above the noise floor.

[@TechnoBabble]

It's surprisingly good. Fast rolling shutter in APS-C for a non-stacked sensor, and it allows for proper 4k capture when used with any of the massive catalog of Super35 cinema lenses (and APS-C stills lenses), something the A7S III/FX3 cannot do.

[@jdohg7594]

I've typically read that it's not pixel size that determines noise performance, but the size/surface area of the pixel that helps it take in light. Of course they are similar, but that could be taken the wrong way, no? I also feel it's kind of hard to test when there are so many differences with these cameras, ESPECIALLY with the denoising. Perhaps I'm missing something.

[@TechnoBabble]

Correct. The technical term for this is the fill factor, it's the ratio between the light sensitive area of the photosite and the other electronics. Modern sensors that might be backside illuminated or have more advanced microlenses over each photosite means that this has become less of an issue over time. A good example I like to share is the Sony A7S III and A7R IV. 12mp and 61mp respectively. Huge photosite size difference and yet they have nearly identical noise performance, actually slightly lower on the A7R IV, up until around ISO 51,200 where the A7S III catches up and then has less noise at ISO 102,400. Now obviously these are extremely high, borderline unusable ISO values for most types of photography, and the A7S III pulling ahead at those extremes is likely less due to the larger photosites and more due to how other electronics in the camera are designed.

[@kirilpetrov2331]

Exactly the proof I was searching for. This is the reason I'm staying away from A7 IV. Not only that you change perspective and POV with 4k60p crop, but you loose dynamic range and low light performance. Also to all the people that say just use wider APS-C lenses - not really the same, because you lose a lot of DOF with 1.5 crop apeture as well. So I'm happy to keep my old A7 III for photography and buy zv-e1 for some B-Rolls and shallow DOF 4k slow-mos. Thanks!

[@TechnoBabble]

It's certainly not a specialized video camera, but I use it in APS-C 4k60 when shooting wedding videos and I don't really have any issues. I just use fast lenses, both full frame and APS-C. The Sigma APS-C primes are great, the Sony 11mm and 15mm are amazing, and even adapting the old Sigma 18-35 and 50-100 f/1.8 lenses works well.

[@Fecabs1]

Thanks man. Very interesting video especially because I was considering getting a 7C II. I am not sure that I need a full frame so I thought well I will use the crop-mode for video so it's fine. Do you think I'll get the same result with the A7C II ? A significantly worse quality in videos with the crop mode ?

[@Niberspace]

100% gurantee yes, this is not about tech it's about common sense physics. "crop mode" literally means that you're throwing away most of the image. If you throw away 10% of the image you'll have 10% worse quality, if you throw away 50% then 50% worse quality, etc

[@Fecabs1]

@@Niberspace Thanks for the reply. I will make sure I make my mind before I buy anything :)

[@SinagMartine]

We are now in an era where we can rest easy and not worry about using ISO 12,800 as it will still provide you spectacular results T_T thank you for this, Sony!

[@81neuron]

oh boy, good clip!

[@courteoussammy112]

What you're saying is that atomic info is better than aggregate info (that can be computed anyway) when you want more total info. Of course that's true. So yes for a given sensor size you have more info with more pixels than bigger ones. Meaning better details, dynamic range... However it's the opposite for rolling shutter.

[@NoahHollinshead]

Great video! ALL HAIL SONY

[@tiziocaio8657]

Sony is cringe

[@thomasobrienphotography]

Thanks !

[@john_punch]

4:38 Why left image is shaking?

[@TechnoBabble]

That would be a combination of my floors and the tripod I used not being super stable. I likely had my feet too close to the tripod and my floors in that room have a little bit of give in some areas.

[@sjmedia_official]

So is it better to shoot 1080 60 if I want to slow down the Image? Or is the quality even worse?

[@TechnoBabble]

Up to preference. The full frame 1080p recording is line skipped/pixel binned, so the quality isn't great and is going to be worse than the APS-C crop 4k. However, you'll get more background blur and subject separation in full frame 1080p when using similar primes or zooms. I have tested 120fps though, which is 1080p in both full frame and APS-C crop. It's usable in full frame and absolutely terrible in the APS-C crop mode.

[@sjmedia_official]

@@TechnoBabble is there even a point to get a CF express for S&Q mode? Is the quality better in this mode comparing to slow motion HD?

[@TechnoBabble]

If I'm not mistaken the only S&Q mode that requires CFe-A is the 4k60 XAVC S-I, since it's higher bitrate than even the XAVC S-I 4k60 in real-time recording. I would say don't bother. I see no use for CFe-A on the A7 IV, outside of super intense sports or wildlife photography.

[@Niberspace]

@@TechnoBabble Hi TB, why is that? According to the manual it seems both regular 4k60 and S&Q 4k60 is using 600 bitrate, so why is it that only the S&Q mode requires that faster card? if anything the regular mode should be slightly more demanding since it has audio

[@TechnoBabble]

@@Niberspace Your guess is as good as mine. Even weirder, the XAVC S-I S&Q 60fps to 24fps (240mbps * 2.5) requires CFe-A but the 60fps to 60fps (600mbps) also in S&Q only requires a v90 card, like the regular recording mode. In both of these modes the overall bitrate of the file being written to the card is the same.

[@mirrorlessmemory]

I wonder why Josh Sattin’s low light results show the FX30 with more noise in the shadows than the a7IV in crop mode. Did he do something wrong in his tests?

[@TechnoBabble]

Haven't seen it myself, but I'll check it out. I do know that the FX30 does a bit less denoising than Sony's other cameras, but that can be a good thing for detail. It also could come down to a slight dynamic range difference, depending on how the test was done, since the basis for the FX30 sensor is an older technology than the sensor in the A7 IV.

[@TechnoBabble]

Update: I checked out Josh's channel, but you're gonna have to point me to a video where he showed a low light test between the FX30 and the A7 IV in crop mode. All I found were multiple videos where he compared the FX30 to the A7 IV in full frame recording, where obviously the A7 IV has better performance.

[@MrPhillipsTheChangeAgent]

Are you not suppose to expose to 1 stop over?

[@TechnoBabble]

While it is generally good practice to ETTR (Expose to the right of the histogram), I'm not really sure what that has to do with the purpose of the video or my findings.

[@duyhungphannguyen3049]

how about the crop mode in A7C bro? is it the same as A74?

[@TechnoBabble]

The regular A7C? No. The APS-C crop mode for video on the A7C and A7 III is very poor quality and should basically never be used.

[@duyhungphannguyen3049]

@@TechnoBabble can you explain a little bit more why? Should i usse clear image zoom instead of crop mode? I am thinking about getting an APSC f1.2 lens to use on my a7c :D

[@TechnoBabble]

@@duyhungphannguyen3049 You should probably test it for yourself, but I believe clear image zoom will give a better result than APS-C crop. Also, something to note, an f/1.2 lens for APS-C will give about the same results as an f/1.8 lens for full frame. Plus, because of the poor APS-C crop quality/having to use clear image zoom, you'll probably get a better result with a cheap full frame lens instead.

[@winfilms3681]

crop mode has substantially better rolling shutter performance

[@TechnoBabble]

Correct, but that wasn't the subject of the video.

[@Niberspace]

In other words, I guess this proves that A7iv simply is terrible for HFR (High frame rate) stuff: 4k@60 crop is bad, 1080p60 has pixel binning. 1080p120 has no audio (S&Q mode only). I'm kind of embarrased spending a near fortune on this camera which can't even do >30p (heck, even 30p it can't do in H265 mode). Kind of embarrasing that any kid with a smartphone can take better sports footage than me with my >10k$ camera investment. I think the solution is to simply focus on LFR for now, but since slowmo really does make stuff look cinematic I sometimes wonder if I should just cut my losses and sell the cursed camera for a big loss.

[@TechnoBabble]

That's very hyperbolic. Nearly any smartphone's video is going to look worse than the binned 1080p120 from the A7 IV, especially in any non-ideal conditions for the smartphone... like zooming at all or shooting in any lighting except for bright daylight. I've used the crop mode for 4k at 24, 30, and 60 fps. Just cause it's not as good as the full frame recording mode doesn't mean it's bad. I've shot commercials and weddings with it plenty of times and never had a complaint about the quality, nor have I personally had any issue with the quality of the footage when grading.

[@Niberspace]

@@TechnoBabble have you used zoom on modern smartphones lately? they zoom insanely well (using 5 different lenses for various zoom levels) But yeah the quality itself is still ofc better on A7IV despite the flaws, but what I meant was I added to the equation all the other benefits that smartphones has, primarily that they are always ready and you can put them (and gopros) in really convinient locations (whereas I don't dare to leave my 5k$ camera setup next to the basketball hoop). So to justify the inconvinience of a big/expensive camera I try to aim for ideal quality (which the crop or binning doesn't allow for)

[@TechnoBabble]

@@Niberspace It's not even remotely close, though. The best smartphones are almost comparable to mid range cameras in video... if you're in perfectly ideal lighting conditions for the smartphone, you're comparing them at wide angle, and you manipulate the test to be favourable to the smartphone. As soon as you move away from those ideal conditions for the smartphone, there's no comparison at all. You're being extremely hyperbolic about the loss of quality when using the APS-C crop mode. Compared to cameras from just 5 years ago, it's still incredible. All I'll say is that If people were able to use the 5D II for TV shows and insert shots/crash cams in blockbuster Hollywood movies then the problem isn't the camera if you can't get looking footage out of the A7 IV in crop mode.

[@Niberspace]

​@@TechnoBabble So the reason I specified "Sports footage" is because I'm thinking of my specific daily usecases which are strangely enough probably what you'd call "ideal lighting conditions", technically I could play soccer/bball/golf/kayak at night but I tend to do those things when it's sunniest midday on open fields. Certainly I wouldn't use my S22 Ultra for my cinematic shots in dim environments (tho sadly I don't get much chance to do that)

[@CaleMcCollough]

You are 100% wrong and comparing apples to oranges. If you center crop the FX30 with clear image zoom and compare that to the A7 IV clear image zoom center crop, the A7 IV is lower noise with better color accuracy and more dynamic range because it has larger solar panels in the sensor. The 24MP A7C is also better at low light than the A7C II because the pixels are better, and you really notice it past ISO 10,000. You don't get good low light without downsampling until you get to the 12.1MP 3:2 range like the FX3/A7S II sensor. What kind of de-noise settings did you use in post for the APS-C mode comparison? I would like to see a comparison of ZV-E1/A7S III/FX3 to the A7 IV/A7C II in APS-C model with de-noising in post to make the de-noising the same. Just because there is noise doesn't mean it's bad. You need to compare with low, medium, and high denoising settings to look at softness/sharpness. Noise is only one metric, there is also color accuracy. In APS-C mode the A7 IV and A7C II both have 4K DCI horizontal pixels (4096) but Sony lists the MP in 3:2, not 16:9, so you have to multiply 4096*(9/16), which is 2304 vertical pixels. Doing the math of 4096*2304 / (1920*1080) you get 4.6K or 9.4MP. You are not down sampling the 4K image much, especially with lens correction on, it's only 4.6K in 16:9 mode, so you would expect more noise because it's not full frame. For me I want a softer image, not a sharper image because I have an FX30 & A6700 for that kind of image. I am using APS-C cinema lenses on on the 12.1MP 3:2 ZV-E1 zoomed in 1.5x with clear image zoom, which is kind of like a 2K image upsampled to 4K but not because Sony combines two pixels to make one due to pixel geometry so it's kind of like 2K upscaled with some DSP so the image is softer, which is good for use with 7Artisans T1.05 cinema lenses to get a look like an old Western or the original Charlie and the Chocolate Factory.

[@TechnoBabble]

@@CaleMcCollough There was no post noise reduction used on any of the clips, as that would defeat the purpose of the comparison. The A7 IV and A7C II are both downsampling to get 4k in their APS-C crop mode. Their APS-C crop mode sensor area is 4600 pixels wide, not 4096.

[@CaleMcCollough]

@@TechnoBabble Look at Sony's website and you'll find the A7 IV and A7C II are both 4096 horizontal pixels (i.e. 4K DCI) (4.6K) in APS-C mode, so there is no significant down-sampling in APS-C crop mode, just some extra pixels for lens correction, and that is why it's noisy, but it's much less noisy that that FX3/FX6/A7S II/ZV-E1. You're going to lose about 2 stops of dynamic range from not down-scalling. This is why bigger pixels actually matter for video, just not for photography. For me, the softer image of the A7 IV is preferred, but the ZV-E1 at 1.5x clear image zoom is 2K and that looks perfectly soft when upscaled to 4K.

[@TechnoBabble]

@@CaleMcCollough Unfortunately you have some fundamental misunderstandings about how cameras work and what causes noise and what effects dynamic range. You don't just magically lose dynamic range on cameras that don't downscale, and ESPECIALLY not 2 stops. Not only that but you're also just contradicting yourself now. 4096 is not 4.6k. Those cameras also do not even record DCI 4k, only UHD. So they're downscaling from 4.6k to 3840 on the width.

[@CaleMcCollough]

@@TechnoBabble I'm a computer engineer. I could build that sensor from scratch. The down scallikng filter on Sony cameras adds a little bit more than 2 stops actually. It's 13 stops without the downs-calling filter, and 15 stops with it. So you're trying to claim the dynamic range does down when you downscale? Why would higher megapixel cameras have "better low light"; it's not true. Better low light can mean more sharp or less noise. Clearly the full frame sensor has less noise. Why did I just buy the ZV-E1 if the larger pixels didn't help? Beware of KZbinrs who make technical videos about science they barely understand.

[@TechnoBabble]

@@CaleMcCollough A "computer engineer" wouldn't know the first thing about how camera sensors work or how to design one... This has to be a troll. Let me know how you testest the A7 IV without downscaling and I'll believe you. 🤣 And I don't mean the full frame, binned 1080p, that's not part of the equation here. I mean 7k full frame or 4.6k APS-C, as in without any downscaling.