jcs

Andrew, the math has been explained that even though only the luma could possibly be 10-bits from an 8-bit 4K to 1080 downsample, in practice all that's going to happen is a low-pass filter and any additional luma variance will come from noise; not the same thing as capturing and directly storing 10-bits from the sensor. From the examples shown, one advantage of the A7S over the GH4 is more dynamic range from the sensor. Folks have already tested the GH4 10-bit external output against the 8-bit Canon Cx00 and the GH4 is significantly outclassed for post latitude. 10-bits can help with tonality gradations, however if true sensor dynamic range wasn't captured, an 8-bit system with more real sensor DR will outclass it. If this explanation doesn't make sense, you can do the tests yourself with actual footage. Thomas Worth wasn't enable to show there's real 10-bit gradations with his own custom software. If you want to be fair and accurate, you might want to revise the GH4 10-bit luma statement until you prove it to yourself one way or another with actual footage.

Shutter was either 1/25 or 1/50 (tested many settings- doesn't appear to be in the exif data?), ISO 1600. Unless it's a zombie, alien, mutant or similar movie shoot, makeup matches skin tones :)

GH4 + Voigtlander 25mm @ F.95, very low light. 1920x1080: '>

Does your preamp supply phantom power? If so, you might try it instead of the mic battery (might sound better). If it doesn't sound better, you'll get more runtime from the preamp using the battery on the mic.

If using the VMP you could skip a preamp and try +20dB on the VMP and -12dB camera mic gain. If you don't have a VMP I'd go with a shotgun with XLR input and phantom power and use a preamp: higher quality and more flexibility.

Right- Canon skips debayering altogether on the C100/C300 by averaging two green Bayer grids and taking red and blue as is from a 4K Bayer array. That said, the 1920x1080 RAW Bayer array debayered with simple+fast bilinear (mlrawviewer) still looks much better than what comes out after Canon's H.264. It looks like perhaps they are adding a Gaussian/box blur before the compressor stage. The good news is that 5D3 H.264 sharpens reasonably well in post, though the color isn't of course as good as 14-bit RAW, especially if procesed with ACR.

The Rode VMP has the buzz; not clear if the DMW-MS2 truly has no buzz as reviews state it's very noisy (and not very directional). A solution is to use an iRig Pre as a preamp. More info here: '?do=embed' frameborder='0' data-embedContent>>

I analyzed the GH4 audio buzz issue yesterday. While I couldn't find any issue with the internal mics, the external circuit is picking up camera electronics, primarily the electronic shutter (buzz varies predominantly with shutter speed). I tried a few mics with TS, TRS, and TRRS plugs. The problem is not with the plug itself- the buzz is the same with them all. If the Panasonic DMW-MS2 with TRRS plug really works without the buzz, then there's something going on with resistance and/or capacitance/grounding (or more). Reviews of the DMW-MS2 appear to show that the mic is both noisy and not very directional. It's a mid-side mic however decoding is only available on the mic (and controlled with camera settings). I use an Audio Technica BP4029 stereo shotgun on the FS700 (mid-side) and it's great being able to control the stereo field in post varying from pure directional shotgun (mono) to a wide stereo sound stage, and everything in between. The stereo is so good that listening with headphones on you'll turn around sometimes thinking the sound is live- it can be that 3D (similar to binaural). The current workaround is going back to the standard DSLR method: setting in-camera gain to the lowest and using a preamp. For the GH4 that means setting mic gain to -12dB. At -12dB, the buzz is still present, however when testing with a Sennheiser G3 wireless system starting at -30dB receiver output and going up from there, the buzz seems to take a sharp reduction at around around -18 to -12dB G3 receiver output. Perhaps the GH4 has some form of AGC going on? In any case, at around -12dB (or higher) G3 receiver output, the SNR is high enough that examining the waveform in post shows the buzz is inaudible and audio is then usable. When testing with a Rode NTG-2 on battery with a TRS connector, the noise floor (white noise) isn't that bad on the GH4 (compared to the 5D3 for example). However the motor-boat buzz renders the audio unusable. The buzz is a triangle wave, about 2x the amplitude of the noise floor. More info and waveforms + spectral plots here: http://www.dvxuser.com/V6/showthread.php?324401-External-Microphone-Solutions-for-the-GH4-(avoiding-the-buzz) A low-cost, reasonably small and lightweight solution is an iRig Pre, for about $33. Along with a splitter cable (or with the soldering mod) it is small enough to attach to the back of a shotgun mic with Velcro or a rubber band. With a 9v battery, it supplies phantom power and sufficient gain for clean, high-quality audio.

I did shoot video for the background- time was an issue and removing cars, people, parking meters, etc., is slightly easier for one frame in Photoshop ;)

Our first shoot with the GH4 in 4K, edited and delivered in 4K (everything played in real-time in PPro CC ("Full") during editing; 2010 12-core 24GB MacPro and Quadro 5000). Watch in full screen if using a >HD monitor (1440p and 2160p available): Audio recorded with the FS700 (Audio Technica BP4029 stereo shotgun (Mid+Side)). Close ups with FS700. Background plate shot with GH4, JPG, full auto. Keyed with Ultra Key. Since many of the items were also green, I had to perform a bit of post work using Track Matte keys to mask off areas from the green screen keyer. A couple areas with motion not masked off and those go transparent (can be fixed by animating a Track Matte key). Could use a bit more color work- ran out of time. Everything done 100% in PPro CC- no round-tripping. File rendered to a custom profile at 3840x2160 32-40Mbit/s 2-pass H.264 MP4.

The FS700 has been challenging for skintones, however after tweaking profiles it's not so bad if carefully lit, white balanced, and exposed for skintones. The GH4 is also a bit thin in the skintone domain. The reason 5D3 RAW is called the baby Alexa is that skintones are handled extremely well, without having to go to extreme lengths to make skin look good. I'm going to the trouble to experiment with the GH4 to find out how to best get good skintones as it is a much more efficient camera to use in post vs. 5D3 RAW. It's also more stealthy when shooting in public- doesn't draw any attention. For wide, outdoor/landscape shots, the GH4 4K is excellent and even the FS700 can't match it for 1080p (GH4 4K => 1080p). Different tools at different price points with different amounts of work and storage: no one tool for all purposes (yet!).

The Canon C100 (and Sony FS700) 24Mbit/s H.264 codecs are on par with the GH4. It's not clear if the 50Mbit/s GH4 1080p can provide better quality vs. the 24Mbit/s codecs, however the image quality won't be as good as the other two cameras can produce higher quality at 1080p than the GH4 (GH4 needs to shoot 4K and downscale in post to match and possibly exceed those cameras (but won't be by much: the C100 is near the Nyquist limit for 1080p, the FS700 just a bit less). In my tests to far, it looks like the 24Mbit/s FS700 codec is doing a better job with color (and certainly dynamic range (14 stops for FS700 vs. 10-11.x for GH4)) than the 100Mbit/s GH4 4K (which provides more luma-edge resolution). The Canon DSLRs are limited by low resolution images coming into the compressor (5D3), and aliasing (most of the other Canon DSLRs). Coupled with low CPU/ASIC power to perform high quality compression with H.264 and not-as-sophisticated-noise-reduction (as with the GH4), the final result from Canon DSLRs is comparatively low compared to cameras with better pre-compression frames and more powerful+modern compression and NR. 14-bit 11.x DR 5D3 RAW slays all these camera for final image quality, especially skin tones, at the cost of huge disk space and lots of extra work. If Canon were to release a new camera with modern CPU+ASICs and C100 level H.264 or better, they'd easily be competitive again. Apparently Sony is going to give it a go with the A7S (only issue noted so far is RS, which probably won't be any worse than the 5D3).

We'll have the film look from digital cameras once someone spends the time to fully simulate what film does with light. ARRI is already pretty close, as they developed a camera system after many years of experience with film scanners. We'll go way beyond film with real-time HDR + tone-mapping in-camera. Any film stock will be available as a preset, with full control of grain simulations and artful defects/aberrations as well. We're pretty close to film DR with the Alexa, and perhaps have matched it with the Red Dragon. All that's left is the color, grain, and aberration simulation. It may take a while, but it will happen.

Cool thanks!

Looks excellent! What camera settings? (other than UHD and 422 10-bit enabled over HDMI)

Matt- I've watched quite a few GH4 videos, including comparisons to the C100 and 5D3. When properly lit, the GH4 does pretty well. I think the reason there's been so many less impressive videos with the GH4 is improper lighting and/or incorrect camera settings for the conditions. In the case of more advanced users posting videos that don't look great (yet): there's a learning curve with a new camera. There are a few excellent GH4 videos out there: It's also a great stills camera: During our initial GH4 video test shoot, Elena wanted to try some stills: This shot is in full auto, with Elena holding a reflector with the Panasonic 35-100 F2.8 lens, JPG, straight from camera (5D3 70-200 F5.6 equivalent). The GH4 has some interesting features, such as iDynamic and iResolution, which can do helpful things in the right conditions and can look strange in others. Thus, as we learn to use this new camera, the results will get better over time. When set up right for the conditions, the GH4 color science is pretty good, better than the Sony FS700 (the A7S demos look better than the FS700 as well (internal AVCHD). The FS700 with the 7Q is a totally different camera). Canon's color science appears to do better when things aren't perfectly set up, and allows for more post latitude (RAW). This is one of the reasons the ARRI Alexa (and now Amira) are so popular: they create great images with less effort, even when not set up perfectly for the conditions. 5D3 RAW is my preferred still camera and the 70-200 F2.8 II for modeling shots, however it's a massive, attention drawing set up. The tiny GH4 (with built-in flash) with the tiny 35-100 F2.8 can be taken more places, draws less attention, and takes great photos.

Nice review Andrew! The GH4 deserves the positive feedback after Panasonic listened to filmmakers and delivered what was asked for at a reasonable price! I've been looking for a camera to supplement or replace the 5D3 for video for a long time. The first attempt was the FS700 + SpeedBooster. The idea was find a camera which has true 1920x1080 resolution and full frame compatibility with my Canon lenses. At the time I had not studied the math and physics and wanted to maintain the mystical and oft-hyped FF "look". You mentioned " 5D Mark II and III raw from full frame sensor a formidable cinematic look vs the smaller sensor in the GH4". The "formidable cinematic look" is a myth. Perhaps a better way to say is that the full frame format has more affordable and flexible options for shallow depth of field- as that is all it really offers in absolute terms of sensor size. Canon's color science is still a formidable challenge. In low light, the GH4 has a nicer noise pattern. 5D3 RAW has some fixed pattern noise and the general noise itself gets downright ugly in low light. The newer GH4 does a better job with noise once it shows up. If I find the time I'll post tests comparing the 5D3 RAW with the Canon 50mm 1.4 at F2 to the GH4 with a Voigtlander 25mm F.95 at .95 (same as the 5D3 at 50mm F1.9 per the math here (which isn't anything new; I finally took a look at it after the Northrup video was posted on EOSHD: '?do=embed' frameborder='0' data-embedContent>> ). The GH4 looks fantastic straight out of the camera when set up and lit correctly. This is something I find valuable and useful. 14-bit 5D3 RAW offers far more options in post, especially when using ACR vs. the comparatively fragile GH4. The RAW workflow might be called "Really Awful Workflow" due to the extra time and storage requirements. That said, for some shots and projects, the extra work is worth it. Regarding GH4 10-bit luma and 444 for 4K scaled to 1080p: the 444 is real, the 10-bit luma is not in practice. 10-bit color processing doesn't take place in the GH4 unless 10-bit 422 is enabled for HDMI output, otherwise processing for video is 8-bit. In order for the 10-bit luma to be real, we'd have to do special encoding in the camera which we could then recover in post (while reducing luma resolution 4x). Otherwise we're just getting a nice, useful, low-pass filtered, noise reduced luma in the same way as the C100/C300. When comparing to the Sony A7S- the example A7S videos appear to produce excellent, detailed, color accurate, high-dynamic range footage straight from the camera. The bitrate is only 50Mbit/s, however as we've seen with 100Mbit/s GH4 4K, which is 4x the pixel data(!), these bitrates are OK as long as motion is smooth. Even the 24Mbit/s FS700 footage compares very well when the motion is smooth. As many of us have noted, even with large, cheap storage, efficiency is important in the long haul. As I'm sure future tests will show, 50Mbit/s A7S in-camera 1080p will compare very well to the GH4 4K scaled to 1080p in post. The A7S appears to have a dynamic range advantage, and the color science so far is very competitive, even against 5D3 RAW. The GH4 with fast micro 4/3 lenses has a size/portability/stealth advantage. In the event 4K material is needed (vs. 1080p scaled in post), the GH4 has the advantage (including reframing in post). Neither camera 'wins'- they have complementary features. The A7S compared to the 5D3 RAW- that's another matter: the A7S will best 5D3 RAW in resolution, detail, dynamic range, and workflow. What remains to be seen is final color science (rolling shutter will likely be a wash between the two).

I watched 5D3 RAW upscaled to 4K vs. ARRI Alexa, Red Dragon 6K, GH4 4K, and Canon 1DC 4K, on a ~20' screen with a Christie 4K projector (Hot Rod Cameras GH4 event at Blacklist Studios in Hollywood): 5D3 RAW looked excellent. The other cameras were sharper, however the 5D3 RAW could easily be cut between them. I recently picked up the GH4 and it compares very well to 5D3 RAW and provides even better resolution than the Sony FS700 (haven't compared slomo resolution yet), which provides higher resolution than 5D3 RAW. The GH4 is more detailed, however the 5D3 14-bit RAW color is comparatively amazing, especially if processed in Adobe Camera Raw (ACR). If you're shooting for the big screen, the extra work for 5D3 RAW will be worth it. It is the best value for getting Alexa-like imagery for low cost. You already have a 5D3, it's just the time to download and install ML to test for yourself. The GH4 will require really fast lenses to get the same FOV and DOF vs. the 5D3 (I picked up the Voigtlander 25mm F.95 to get the 50mm F1.9 5D3 equivalent. They also make a 35 and 85mm F1.9 full frame equivalent (17.5 and 42.5mm F.95): each are >$1k and full manual (not a big deal for video). Upscaled 5D3 2K to 4K won't look any different than 2K 5D3 RAW projected/shown at 2K. Thus GH4 4K will always be sharper and more detailed. However, the color processing capabilities of 5D3 14-bit RAW are far beyond what is possible with the GH4 4K. The GH4 4K can produce amazing footage straight from camera with little or no post-processing, which saves (a lot of) time and disk space vs. RAW. If you want to get the most color and detail from 5D3 RAW, use Adobe ACR. Resolve is a close second and if using sharp lenses and shooting outdoors, the results are very impressive:

yiomo- I was going to wait for the A7S, but after realizing that the "FF advantage" is a myth, I purchased a GH4. Time will tell if the A7S is a better camera, however it will be because of the sensor technology, image processing, and color science and not the sensor size (more info here: '?do=embed' frameborder='0' data-embedContent>> ).

The Northrup video (thanks KarimNassar for posting) got me thinking about the following math to match a crop sensor to an FF sensor camera (this was also helpful: http://www.josephjamesphotography.com/equivalence/#introduction ): Multiply the focal length by the crop factor Multiply the f-stop by the crop factor Divide the ISO by the square of the crop factor The first two equations deal with the difference in scale. The physical aperture (light opening, "entrance pupil") will be exactly the same: Given: f-stop = focal-length/aperture-diameter aperture-diameter = focal-length/f-stop GH4 Lens: 25mm f2.8 5D3 Lens: 50mm f5.6 GH4 aperture-diameter = 25mm/2.8 = 8.93mm 5D3 aperture-diameter = 50/5.6 = 8.93mm Thus, the aperture diameter or entrance pupil will let in exactly the same amount of light. The 5D3 sensor has a focal length that is 2x farther from the entrance pupil, and since the sensor is 2x bigger, it captures the larger projection of exactly the same amount of photons. Since we've spread the projection of photons out with a 2x larger area, the sensels will get 2*2 (area) = 4 times less light than a 2x crop sensor. So, we have to boost sensor gain 4x to match the crop sensor. Now, if due to manufacturing or technology advantages the full frame sensor is more sensitive per sensel vs. the crop sensor, then a 4x gain boost to match cameras won't be accurate. FOV and DOF will be exactly the same regardless of sensor technologies. The only differences will be sensitivity, noise, and color characteristics. After starting with a 5D Mark II, then going to a 5D Mark III, then adding a FS700+SpeedBooster, I began to suspect that there was nothing inherently special about FF. After reading about and understanding this math and physics, it's clear that there is no mathematical or physics-based advantage (in terms of light and photons) to FF over smaller sensors. Only when a larger sensor can be made more sensitive, less noisy, and/or provide improved color processing can a FF sensor perform better than a crop sensor. The main reason the 5D2/5D3 became so popular was due to Canon's superior color processing. The 5D3's softness combined with low aliasing and excellent color processing for skin tones helped make it very popular as that is similar behavior to film. The ARRI Alexa has the best color processing and until the Dragon was released, the most dynamic range (not clear yet if the Dragon has matched or passed the Alexa, however it does an excellent job with skintones). If we consider the SpeedBooster, using the same lens on both cameras, so the entrance pupil is the same size and the focal length is the same, then we'll have exactly the same FOV and DOF when the focal reducer shrinks the image circle down to the crop sensor. This means the f-stop is not changed, just the t-stop (the SpeedBooster for Canon EF to NEX is really a 1.1 crop, however the bokeh was nearly the same in testing between the 5D3 and FS700 with the 24-105mm F4L lens). This is one case where the smaller sensor has an advantage: since we're shrinking the photon projection area, we are increasing the amount of light to the sensor (thus gaining the ~1 stop of light increase in the EF to NEX case). A current market advantage for FF over 2x crop is that there are more fast lenses for FF. We had to get a Voigtlander 25mm F.95 to come close to a 50mm F1.4 on FF. We'd need a 25mm F.7 to match the 50mm F1.4's entrance pupil size and thus bokeh. To match the 50mm 1.2, we'd need a 25mm F.6! Other than available/affordable lens choices, sensor technologies and color science, there's no mathematical or physical advantage to FF over crop sensor cameras. The noise characteristics of the GH4 are finer and nicer than the 5D3 and FS700 in low-light testing so far. The detail captured by the GH4 exceeds the FS700 (and slightly passes the C100/C300 (for 4K downsampled to 1080p)). 5D3 RAW still has the nicest skintones, however we're still learning to use the GH4. Understanding that there's nothing magical about FF got me more interested in giving a 2x crop sensor camera a try- so far the quality is impressive.

http://nofilmschool.com/2014/05/6k-red-dragon-35mm-film-canon-5d3-magic-lantern-raw-atlantis/#more-82041 I'm currently testing the GH4 against 5D3 RAW and the FS700. Initial impression is how to soften some shots for the GH4- the opposite of the 5D3 H.264 (too soft). 5D3 RAW isn't full 1080p (which is what downsampled GH4 4K to 1080p represents (hits the Nyquist limit)), however it's just about right for actor close-ups. The FS700 AVCHD is almost full 1080p, but has some aliasing artifacts and 'black edges' at super-white boundaries (though has 14 stops DR and excellent slomo up to 240fps (not full 1080p + more aliasing (still excellent for what it is)). In the video test above, they test various softening filters for the Red Dragon at 6K. The 5D3 RAW footage will easily cut between film and softened Dragon 6K footage. I will test softening the 4K GH4 before downsampling to 1080p to see how well it can match 5D3 RAW (the GH4 has a little aliasing at 1080p native; blurred 4K before downsampling will be as alias free as possible. Mathematically, that's the correct way to downsample- low-pass filter (blur) then downsample. It's not clear if it will provide a visible difference vs. blurring after downsampling if the source material is already alias free (and the downsampling algorithm doesn't introduce aliasing)).

http://pro.sony-asia.com/pressrelease/asset/570730/section/broadcastreleases (states 8-bit XAVC-S is 420; 10-bit is 422). http://***URL removed***/news/2014/04/06/sony-announces-alpha-7-series-full-frame-mirrorless-with-4-2-2-4k-video-output (420 for internal 8-bit 1080p recording)

A7S does 420 internally: 422 externally.

I might be able to shoot something this afternoon- will post in this thread if I do.

The concepts from the video and the math are helpful. A GH4 with a Voigtlander 25mm F.95 at ISO 200 will look the same as a 50mm F1.9 ISO800 on a 5D3, with about the same amount of noise. The same scene and settings requiring ISO6400 on the 5D3 will need ISO1600 on the GH4: perhaps similar low-light performance (Northrup cleaned up GH4 ISO12800 footage with Neat Video (pool shot in 5D3+GH4 video): looked decent on youtube).