Table of Contents Show
Archive notice: This article was originally published on July 1, 2017. Links and embedded videos are preserved as part of the historical record.
In good light, the almost 12 f-stops of dynamic range of the GH4 were completely sufficient, even in direct sunlight the lower dynamic range against RED or Arri did not matter, provided the exposure was optimised.
Product cycles
Panasonic took its time with the successor, well over two years. On the one hand, this is good, as users are not constantly forced to buy a new camera due to new mini product updates, or at least don’t have to think about it all the time. However, the fact that the competition is running away with other cameras, such as the Sony A7s MK1 or MK2, is always an issue – with BetaSP cameras you easily had 6 to 10 years of peace and quiet and could concentrate on taking beautiful pictures.
As far as the competition for comparable moving image cameras is concerned, let’s keep it short: the A7s is the only serious competitor, around 50% more expensive and a full-frame camera, but unbeaten in terms of low-light sensitivity, although it only shows its full advantages in almost no artificial light at night – you can even shoot fireworks at night with the GH4. The blood moon, however, can only be captured with the A7s.
However, the A7s also has disadvantages, such as lower frame rates, only 8-bit internal and external recording options and significantly lower detail resolution and, in particular (when it comes to stabilisation), poorer internal sensor stabilisation. And the operating functionality? No comparison.
Review vs. reality
Before we get down to the serious review, we need to put various other claims about the GH5 into perspective. “All 4K footage taken using full width of sensor (oversampled from 5.1K footage)” sounds good, but in detail it means the following: Yes, all pixels are used here, as many other cameras have done – for comparison: a Canon 5d MKII always created the HD video by omitting pixels, and the image had a corresponding number of moirés. Analysing all pixels and including them in a video file requires a lot of computing power and memory and therefore also a lot of energy, a problem that Sony cameras in particular have to contend with. Short battery life, overheating during video recording – in the worst case scenario, this can cost you a nice shot or two. This is where the GH4 had one of its great strengths: excellent work in the design of its processing engine and heat dissipation, plus energy efficiency with more pixel processing than any comparable camera.
The running times of the GH5 are dramatically worse than those of the GH4 – as various YouTube reviews claim. I can’t even begin to confirm that. Even at 4k60p, the GH5 is still top compared to the competition.
Resolution issues
When we read in other reviews “oversampled from 5.1k footage” or “shoot 6k photo-video” in order to then “downscale” to 4K, this is also quite nonsense from a technical point of view. The sensor of the GH5 has 5,184 x 3,888 pixels. That’s a total of 20.2 MP at 4:3, and logically less at 16:9.
A 16:9 UHD TV has 24 MP and a 4K DCI projector throws 26 MP onto the screen. Talking about oversampling to 4K is … well, I’ll save myself the trouble of evaluating that again. Nevertheless, this is significantly more resolution than the broad competition has to offer: The Sony A7s has 12 MP, a Sony FS7 has 8 MP, an Arri Amira / Alexa doesn’t even have 6 MP at 16:9 anymore. How sensible it is to upscale this amount into a 26 megapixel file, as the Amira does with ProRes 444 HQ for 4K, is something you should look at on a 4K projector. And at the latest, if you want to zoom in again, the colourist will complain later – no matter how many camera manufacturers sell it as a feature that you can zoom in again.
8 megapixels is only a small overhead to HD (which needs 6.3 megapixels to be usable) – the GH4 already had almost twice as much resolution overhead as an FS7. But it was only really noticed (if at all) by the colour grader, who then had trouble sharpening the lower-resolution camera images so that they matched. If you talk to colour graders about what they have to fix the most, even with HD or 2K films, it’s the lack of sharpness, which is preceded by a lack of resolution. But lack of focus is also the bane of full-frame DSLRs. Another advantage of the GH4/GH5 is that the Micro FourThirds sensor is closer to 16mm and is also less prone to focusing errors.
Another reason for a lack of resolution is the frame rate. In order to bring 4K resolution to the viewer, higher frame rates with shorter exposure times are required. As Arri itself beautifully demonstrates, even at 2K you can already see that 60p conjures up significantly more detail resolution on the screen. With 4K, it is even more pointless to shoot in 24p, 25p or 30p. The motion blur cancels out any resolution above 1 to 2K. And here is the first big plus point: the GH5 is the first photo/video SLR that enables 4K60p recording at a resolution of around 20 megapixels – even internally (8 bit internally, 10 bit externally via the HDMI output).
8 bit vs. 10 bit with V-Log, C-Log, S-Log, Canon-log, Vlog?
Everyone wants to shoot with Log because they believe that everything will be better now – unfortunately, this is also a widespread misconception. Log is a compression method that dates back to the early days of digital workflows, and you’re only casting out devils with the devil with Log. Anyone who is anyone in VFX has been working exclusively in linear and not in log for a long time. I don’t want to open a new can of worms on this subject, but in a nutshell, log is only worthwhile if you have scenes with a lot of dynamic range and can also record with at least 10 bit – 8 bit quickly produces banding, especially if there is little dynamic range.
Since the majority of cameramen also underexpose this format – they obviously love to torment colourists – banding and image noise are a frequent reason for costly repair work during grading, at least with low-cost recording formats such as H.264 in 4:2:0 – 8 bit, but also with 10 bit.
Now the GH5 finally offers 10-bit internal up to 4K30p or 10-bit external at 4K60p, and is the first in the price range to do so. And it is only now that vlog is starting to make sense and that almost 12-bit colour depth can be tipped into the post. However, Vlog is still a chargeable update. But the GH4 already had a trick to add more colour depth to the post: record 4K and scale it to 10-bit 4:4:4 ProRes HQ in HD using AE, for example. By downscaling, you compress the amount of pixels, you have plenty of overhead and the scaler in After Effects is very good. You can feel the difference in the grading. The compression artefacts are also smaller and there is less noise.
Sensor stabilisation, or who still needs a gimbal?
For film shoots, a stable camera (dolly or steadycam) was only affordable for high budgets. All the more reason why gimbals have enhanced low-cost productions and given the masses an image quality that was previously unaffordable. Now, with the GH5 and its Dual IS (Dual Image Stabilisation), we are almost at the same level as a GH4 with a gimbal. Whereas the optical stabilisers in the lenses were mostly only suitable for photos, you could hardly pan with film. In the new model, you no longer feel any hangs even with very spontaneous panning. From a practical point of view, it is now a pleasure to be able to react to actions “just in time” in “Run & Gun” and to be pleased with the stable image in the post afterwards – without having to sacrifice documentary mobility and breathing. All other camera manufacturers can learn a thing or two from this, the feature is currently unrivalled. Here is an example filmed with the 100-300 mm at 300 mm freehand with Dual IS in 4K50p slowed down to 50%, i.e. 4K25p.
Autofocus
Anyone who has watched a few YouTube reviews of the GH5 has inevitably come across this hot topic, which has caused a lot of excitement, almost a “battle of the vloggers” – and as is so often the case, a storm in a teacup. My experience was different and also has something to do with the fact that vloggers shoot differently and have different demands than a real cameraman.
Vloggers want to hold a black product packaging 10 centimetres in front of the camera and dream that it will be in focus immediately.
I (like many cameramen) used to classify autofocus as unsuitable for video and therefore even had certain reservations about taking such a feature seriously at all. It wasn’t until the Sony a6300 came out that autofocus became usable – although you can see quite a few vloggers driving in the car or walking around blurred. As part of HandsOnXk, I dared to use autofocus for the first time to shoot a quick making-of – and was surprised at how well this was utilised in practice. The new joystick is a great help, allowing you to quickly move the autofocus field (e.g. when two people are facing each other and no one is in the centre).
Speedbooster: film look and playing with the blur
Nevertheless, I am quite a trained manual focus puller, and if you prefer the film look and want to hide the background with as much blur as possible, then you can do this with an M43 sensor using a lens regainer or speed booster. If you adapt a full-frame lens to microfourthirds using an additional lens package, the larger light circle of a Canon EF lens, for example, is compressed to the smaller microfourthirds circle, resulting in a light gain of around one f-stop, of course in relation to the full-frame format.
The Canon f1.4 50 mm becomes an f1.0 75 mm if you also convert the crop factor – this results in a great look. I use the Apurtur Lensregain – this has the advantage that you also get a radio remote control, which you could also mount on a gimbal, for example.
This allows you to quickly programme two focus points and then move them precisely. The GH5 also offers such a feature in its own menus, but it’s not too easy to access. This would be more suitable for working with a slider, where you take more time for the settings anyway.
400 Mbit All-Intra compression is still to come
One feature that will finally elevate the GH5 to the throne of independent cameras will be an IntraCodec suitable for post-production at 400 Mbit bitrate, which is due to arrive in the summer. Why is this function not yet being released?
Technically, the camera would be able to do this. The performance is there, but until recently the card manufacturers have not yet brought any cards onto the market that deliver sufficient performance. This topic has also been hotly debated in various forums, as 400 Mbit is “only” around 50 megabyte/s. The Sandisk cards that were considered to be ideal for the GH4 have had a performance rating of around 280 Mbyte/s on the card for over a year?
Well, as is the case with technical specifications: if no details are published about what is meant by this or under what conditions this can be achieved at all, the buyer can only dream about what is possible or not. In fact, this refers to the possible peak and this is only achieved if you want to save individual images in short bursts, for example when shooting continuous images. However, if you want to record film constantly, you will be severely disappointed that this is only possible at just under 30 Mbyte/s with a 280 MB/s peak SD card.
A new standard has therefore been defined for SD cards, which now provides clearer information as a label about what is possible as a constant write rate for film: “Vxx”. V stands for video and xx for 30, for example, so “V30” would be around 30 Mbyte/s. The 280 MB/S card is therefore actually a V30 card. however, 400 Mbit requires 50 Mbyte/s, i.e. at least a V60 card. Such cards first have to be produced and then tested and qualified by Panasonic and the firmware may have to be adapted again – and the driver updates for the card reader are still to come. So we have to be patient, the first V60 cards have only just appeared on the market.
More cool features: Focus braking, 180 fps slomo
While exposure bracketing for more dynamic range, i.e. HDR, was previously known in the field of photography, the same is now also available for focus. A feature to generate unusual textures in the macro area, for example. If you are not sure whether the ears or the eye are in focus, the camera can now continue to focus around the focus point and you can then search for the correct focus point from the images in post – no more blurred images.
Another advantage of this technique is that you can use the same technique for macro photography, for example, to merge the images with the focus point in post, similar to the HDR principle, in order to define a larger focus area than would have been possible with the lens/aperture combination.
Stacking workflow
First go to File > Automate in Photoshop and then to Photomerge. This opens the interface in which we load our images in sequence, with the layout suggestion set to “Auto”. It is important here to uncheck the “Blend Images Together” box.
This creates as many layers as images have been loaded and aligns them automatically, which is useful if you are taking photos handheld and the camera has moved, or to compensate for optical pumping when shifting the focus point. Then select all the layers and choose the auto blend function.
180 fps Slomo
The GH4 was capable of a maximum of 96 fps slow motion and the image looked more like upsampled and correspondingly pixelated SD. The GH5’s 180 fps looks drastically better in comparison, and the image is also mostly watchable in HD – I’d rate it at a good 720p resolution. If you sharpen it up a little in post, most normal viewers won’t notice the lack of resolution, but a professional will. However, apart from Arri, no sensor manufacturer can achieve 180 fps at full sensor resolution, even the REDs usually stop at 120 fps. With the GH5, the 20 megapixels are still possible at 4K60p. Compared to a 24p YouTube stream, this would be extremely respectable at almost three times the slow motion. If you don’t believe it, you can already find some good examples on YouTube.
Professional audio support as an optional extension
Panasonic has implemented the audio support as a hot shoe adapter, which is a simpler solution than the huge adapter of the GH4, which also provided audio connections in addition to SDI. This offers two balanced XLR connections on the rear. The picture below shows the complete audio functions familiar from broadcast cameras. No DSLR or DSLM has ever been provided with more audio support.
Conclusion
This review has not covered all the new features, only those that are of most interest to the film/TV industry. From my personal perspective, the GH5 is a must-have for anyone who needs something for filming and photography. A 5D Mk III can no longer keep up with it – you can see a corresponding number of Canon switchers in the GH4/GH5 forums.
If you’re also looking for a good timelapse camera, the GH4 was already perfect at this. The function to calculate a Quicktime movie directly in the camera is particularly great – that’s why I already have two GH4s: one for filming and the second for running timelapses on the side. Now there’s the GH5, so I can run two GH4s in parallel for timelapse and zoom-in effects.
The price-performance ratio is unbeatable. For the rare cases where the A7s offers an advantage, you can rent it almost everywhere, and the A7s III is coming soon. As a Sony buyer, you have to be able to suffer.
