A99 interweb video test part 2 – Sony A99 vs. Sony Nex 5n

***NOTE*** This test was performed with TWO separate copies of the A99.  Yes, I was so disappointed by the video from my first A99, I thought it was defective.  So I returned it and got the same crappy results with A99 #2.  Looking at other A99 videos across the web, its not me or my two copies, it’s the A99.  On to my story…

A99 vs. Nex 5n

I got skewered after posting my results of my A99 vs. EM5 video test with wild claims of bias (I just want good video, don’t care where it comes from), flawed testing methods (focus, push record, what am I missing?), that I rigged it to make the EM5 look better (after spending $6000 on a Sony body and lenses) and so on.  Its all ridiculous.  I put a nice consumer camera (EM5) against Sony’s professional flagship (A99) and the Sony got spanked.

So I figured I’d level the playing field by shooting another test – this time with the A99’s tiny cousin, the Nex5n.  The 5n has a 16 megapixel APS-c sensor, compare to the 24 MP full frame sensor in the A99.  So its Sony vs. Sony – and to lessen the complaints from the whiners, I’m using the exact same lens on both cameras … the stunning Zeiss 24-70/2.8.   I also have the LAEA1 adapter to make the lens work on the Nex5n.  To compensate for the 5n’s crop factor, I shot wider on the A99.  I stopped down the lens on the A99 to equate the depth of field difference between the two sensor sizes.  Everything was shot at a base ISO of 100.  Be sure to select 1080p and watch full screen, or even better on a large TV.

As long as everything is in focus, I’m not really concerned about the rest, but the pixel peeping nerds seem to care so I tried to make everything as close as possible.   The results are not surprising – again the diminutive Nex5n took the A99 to the woodshed.  The softness, moiré and aliasing are again so bad on the A99 its completely unusable.  Fine details are once again mush compared to the 5n.  Its disappointing how bad the results are for the A99.  Absolutely horrible.

Ergonomics on the A99 are a dream.  There are two control wheels – one under the shutter and one by the thumb rest – making it easy to adjust aperture and shutter speed when shooting manually.  Other shooting parameters can easily be adjusted with the joystick on the rear, functions such as focus magnify, AF lock, ISO and more all are easy to find as some buttons are convex shaped and others are concave.  The tilt/swivel LCD is something that should be on every camera, it can be placed on top of the camera for low shots, turned to the side for shoulder mounted shots and tilted down for overhead shots.  I love it.  It’s such a pleasure to shoot with the A99, and the stills results are simply stunning as you can see here.  The video leaves a lot to be desired, and if it can’t best a 5d3, the A99 is going back to Sony.

Some are saying that my results are a by-product of a full frame sensor because the Canon 5d Mark III is also very soft.  That may be the case because full-frame DSLR’s have to hatchet down sensor output in the 22-36mp range depending on your camera to just over 2mp for HD video.  Some also say I’m an idiot.  That may be the case, but the A99 needs glasses – it can’t see.  I’m going to put that to the test in my final shootout with the 5d3.  I’d rather not have to make the switch again as it’s a pain to buy and sell a complete camera system.  I need both stills and video, so my investment has to perform in each medium.  I’d love to see something with great sharpness from the A99, but I haven’t yet and I don’t think I ever will.  If you’re shooting with one and getting good results please let me know.  Until next time, thanks for stopping by.

chris

About these ads

13 thoughts on “A99 interweb video test part 2 – Sony A99 vs. Sony Nex 5n

  1. This is a lot like people jumping up and down about the gh3 and its ‘amazing’ sharpness. It’s far from unusuable From the youtube video, aliasing looks better on the a99 when dealing high contrast semi-horizontal lines but the nex5n looks sharper. The line skipping becomes really obvious on semi-horizontals that have highlights above and below. (makes sense to me) I own two nex5n’s, two a55s, and an a77 that I use for video. I am an avid alpha mount user- and my a900 is the best pure dslr I’ve used to date. (I’ve used them all extensively.) Re. the alpha77 and the 720p thing. I attempted to duplicate the mitros experiment and found it to be totally unmatched by my tests. The a77 is just as sharp as it should be. In my opinion, the nex-5n’s video provides better video than the a77- mostly just because of the noise. But to say the a77 is unusable is ridiculous. I’ve been paid a decent sum to actually use it for video.

    I’d really love to see 24p out via hdmi to the ninja2 or whatever in a direct comparison to compressed. I don’t think it will matter b/c Sony seems to have failed to handle the pixel arrangement properly I personally feel like Sony dumbed down video b/c they didn’t want to compete with their other service lines. The nex-5n certainly impacted my spending choices when it came to a dedicated video camera. So did the wait for the fs700 to come to market and to find clarity about 4k. I don’t think that Sony made the right call with how they seem to have handled vid. on the a99. Make the best product possible. If the a99 wouldn’t have had these issues, I’d have two of them right now. Now, I’ll just keep renting for video shoots and see if they ever find which way they are or are not going to screw me over for being loyal to their systems.

    Overall, I think the blogger her has a valuable and important point that I wish wasn’t so dramatic and overstated. The a99 footage looks very, very good. It’s not drastically worse than 5d3 and it’s not mush. Aliasing would be my major complaint. What’s most sad about Sony’s video let down with the a99 is that they could fix it with firmware and just won’t.

    Dramatic accusations of ‘mush’ and things like that for the mitros fellow to take his complicated but ill fitting math and clusterfuck one FRAME of video using software that is unfit for the task of deciphering his ‘hypothesis’ and then everybody jumps ship because most of the potential purchasers of an a77 wouldn’t even know where to start making films with any camera.

    1. I appreciate your opinion, your comment and your visit, but the 5d3 bests the A99 by a wide margin – specifically because of the rampant aliasing and moire. It also has a far more robust codec and sharpens better in post. As I said, the aliasing/moire the worst I’ve ever seen from any ILC – and I’ve owned the 5d2, Nex5, 5n, 7, GH2, EM5, FS100, A99, Xpro1 and 5d3. Its soft and full of horrible artifacts. The A99 does shallow DOF nice, but if you need detail its complete mush. I live near the water – look at what it does to water in the harbor shots, the 5d3 is so much cleaner and it actually looks like water and not some video game. To make matters worse you’re saddled with the endless waves of moire and jagged lines from all the aliasing. It was the worst camera purchase I’ve ever made – especially because I switched systems after linking the handling of the A99 so much and that spectacular 24-70/2.8 Zeiss. I wish Conrus would hack that lens to Canon. Again, thanks for the words, but the A99 is a great stills camera that shoots poor quality video. As far as the HDMI out, I plugged my A99 into my TV and the HDMI signal is no better – and its still 8-bit 4:2:0 so unlike the new 5d3 update, the color space stays the same.

  2. Thanks for posting the review and having the “balls” to say it like you saw it. I was just about to buy an A99 after having tried the photo mode at my local retailer, and I was very impressed. I’ve used a Nikon D5100 for about a year now [great for me to learn some skills on] and was very happy with its photos and got reasonable video results. The move to the A99 was contemplated to really move up in the stills and get some really great video….but, your review has quashed that.
    Hmmmm….maybe I’ll stay with Nikon as I’ve read the new D5200 is getting much improved stills and video….and costs way less than the A99.

  3. We use 2 A99s for our wedding coverages together with the FS100. We get some pretty nice results..definitely far from being “unusable”. http://vimeo.com/m/64768827

    I edited my own wedding with some MK3 fotage mixed in..resolution wise they are very similar..so it maybe is a sensor size issue. Smaller sensors are just better for video..codec may also be a factor but the fs100 produces such sharp images compared to the A99/77/5N/7 and they have the same 28mbps avchd codec! So I think its more of a debayering issue. Video is just not optimised yet to be captured from full frame.

    1. I’m getting much better results with the 5d3, I’m going to shoot the same stuff and re-edit. The A99 does do shallow DOF stuff well, but wide shots lack detail and the aliasing/moire is so bad it absolutely kills any wide shot. You have the FS100 to take care of that. The 90mbps All-I codec in the 5d3 is far superior and moire/aliasing has been tamed quite a bit. I’m going to use the new Blackmagic Pocket cam – if its ever released.

      1. The Blackmagic Pocket Cam is the dream cam weve all been waiting for!:) I will be buying one specifically for wide angle, deep dof, high detail shooting which the BMCCs just kills for. Unfortunately I cant find a lens that will give a 16-18mm FOV for the meantime yet so that could be a problem ( i love ultrawide for gliding ). Widest right now is the panny 7-14 (21mm eq). That will do for the meantime while qaiting sor Rokinon to do S16 lenses (:D). I agree with the shallow DOF A99 is being good at..high detail,deep focus shots, just like the 5D3 its pretty weak. I also get massive macroblocking for foliage shots on my Nex 5N (I use it for ultrawide steadicam shots). They are all imperfect due to weak debayering from the uncompressed signal..they are capable of nice sharp detailed images as what their stills can show you..its just the video compression algorithm thats weak, or intentionally crippled. Ergonomics wise I completely agree with passion that the A99 is a DREAM! I hope all recent large sensor cameras are like it! If Sony will only put the FS100’s sensor and processor in an A77/99 body with built in grip..even with its fantastic 28mbps avchd coded 1080p 60p..that would already be the dream canera for us evebt filmmakers. Small files..very sharp detailed 1080p.

        1. I preordered the Pocket Cam as soon as it was available – so whenever it arrives I’ll be adding it to the mix. I’m really looking forward to this one. The footage john Brawley posted looks amazing. I have a great lens that I think will really make the Pocket Cam shine. That’s for a future post.

          http://johnbrawley.wordpress.com/

          I use the 5n on a steadicam smoothie with the 18-55. Its great for adapted lenses and it does a nice 60p. Not quite as good as the S&Q on the FS100 (had one of those too), but it cuts well with the 5d3. Thanks for checking my little blog out.

  4. I had a similar experience a while back. We have a Sony a77 — which was Sony’s flagship at the time I bought it. The video was unusable for anything beyond Youtube low-res video (or even Youtube, even mildly cropped). This was not the case with any other camera I’ve used (and we have a whole array of different brands, many much cheaper and older than the a77). After close to a year of shooting/filming, I did a quick analysis:

    http://mitros.org/p/a77/a77_video.html

    The quality was identical 720p video, cropped slightly, upscaled to 1080p. This doesn’t mean Sony does this — this could also come from ridiculously bad post-processing of some form — but zero post-processing would still give better quality than what comes out of the Sony, and the effect on the user is identical: advertised 1080p, actual is less than 720p.

    The reaction I got from the community was much the same as your own — claims of bias, non-sensical excuses from folks who don’t understand how camera sensors and image processing work, personal smears, etc. I was hoping for a technical discussion, but got almost none. It was rather sad.

    1. Here we go again. No other APS-C DSLR camera from the A77 era or before is sharper than the A77. That includes all Canon, Nikon and Pentax bodies from that time. So to say a “whole array” performed better is misleading to say the least. Only dedicated video cameras and a couple µ4/3 cameras would qualify, mostly for having smaller sensors or dedicated video sensors.

      Because all those APS-C DSLR’s (FF too BTW) perform lineskipping and crude methods to reduce moire and aliasing, which costs them quite a bit of resolution.

      Most of the slack you got was because you *DID* claim sony upsampled 720P to 1080P and implied fraud, without evidence.

      1. >Most of the slack you got was because you *DID* claim sony upsampled 720P to 1080P and implied fraud, without evidence.

        I never made this claim. I was very explicit in all cases. I stated I had evidence which was consistent with Sony upsampling from 720p to 1080p. I use precise phrasing, and “consistent with” is very different from “*DID*”

        The scientific process starts with a hypothesis. A hypothesis is a proposed explanation for a phenomenon consistent with existing observations and that can be tested with further experiments. Once you have a hypothesis, you do experiments, and if those experiments are consistent with that hypothesis, it is upgraded to a theory. For most phenomena, there will be many hypotheses. After experimentation, there may still be more than one theory. You do not move into fact until the evidence is really quite overwhelming.

        Stating that something is “consistent with” experiments or observations implies hypothesis or theory, not fact.

      2. One follow-up question. You state: “Because all those APS-C DSLR’s (FF too BTW) perform lineskipping and crude methods to reduce moire and aliasing, which costs them quite a bit of resolution.”

        I explicitly simulated line skipping. This does not explain the observations. Can you please describe the “crude methods to reduce moire and aliasing” you are referring to? In those cases, I am unaware of any explanation which would be consistent with the observations which I have seen.

        A low-pass filter *prior* to sampling would result in both reduced moire and resolution, and give the results shown. The camera gets the data *after* sampling, and so does not have the option of doing this. A blur after sampling will do nothing about moire.

Please be a part of the discussion and comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s