GH5 Game Over, Panasonic

[Sage]

Did you shoot this at Chernobyl? The saturation is nuclear! ;-)

How did you know

A bizarre combo of maximal saturation, very revealing of the slightest flaw
Just finished the Tungsten test as well, and there I see a difference. The book, which is the most far red I've got, has a distinct magenta shift on the Alexa under Tungsten (surprising, as the GH5 appears more accurate here)
As best I can tell, the two have a very different infrared cut off (the Alexa is more permissive), and this is especially evident under tungsten

Alexa LogC

GH5 VLog

 

[Thomas Smet]

Utter nonsense.

NO COMPRESSED CODEC can compare to RAW, and all GH cameras have drastically less dynamic range than Blackmagic cameras.
Real working pros, like a few right here on this forum cut Blackmagic footage in and out of Alexa footage on a daily basis. If that were remotely possible with ANY GH camera they would do it in a heartbeat.

And your statements about 'problems' or 'failures' that you 'heard so widely' about are also ridiculously exaggerated. I somehow am led to doubt whether or not you've ever actually shot with a Blackmagic camera after saying things like this.

We all need to realize a lot of what we are talking about is splitting hairs. Of course raw is better and without a doubt higher dynamic range is better but what exactly does that mean and by how much really is it better? We have a tendency to inflate the benefits of specs a lot more than we should. Doesn't mean its not better but that we tend to make a really big deal about a 1% visually perceived difference.

We keep hearing people talk about this camera is inferior or that codec can't old up but where is the proof? We can all talk until we are blue in the face but pictures speak louder than words. I have seen a lot more tests posted that praise the quality of the GH5 than I have of tests showing it getting crushed to death. Here are a few such examples:

Cinegal did a rather extensive comparison between the GH5S and the Alexa. This isn't to show that the GH5S is 100% as good as the Alexa but that it can hold up very well when compared to the Alexa. I don't think anybody here pretends the GH5S is as good as the Alexa but what matters is how close can one get with an easy to use camera.

https://www.youtube.com/watch?v=S_r3qiGXTlo

Then there is Stephen Fleet who is a VFX Supervisor on TV series like Marvels Iron Fist on Netflix. You don't get any more pixel peeping extreme than a VFX supervisor and he loves the GH5. He uses it to shoot background plates on set. Of course it isn't the main camera used on these shows but he is very impressed with the look of the GH5. Again this isn't to say the GH5 should be used on Jessica Jones season 3 as the main camera but that in the scope of cameras and the overall look here we have a VFX supervisor who is in pretty good touch with the needs of Hollywood and actually earns a solid living in the industry who rather enjoys the look of the GH5. I tend to take his view over random online comments not just because he is a VFX supervisor but because he actually works in the industry.

https://www.youtube.com/watch?v=ewvh2jPAFNw

Stephens intensive nerd comparison of 150mbps vs 400mbps on the GH5.

https://www.youtube.com/watch?v=uXhn-5UILgk


Finally there is the fact that a lot of Hollywood movies don't even shoot raw anymore. Some of the big budget movies will like Avengers but your typical drama, action, rom-com or comedy is likely using ProRes these days. Not just for budget reasons but ease of use and colorists realizing the end result is virtually the same. As long as one doesn't completely jack up the shot and has even close to half a brain when it comes to exposing the shot then ProRes should be more than enough. I always have to ask the question why is it the low and no budget indies seems to think raw is a must when the majority of movies in Hollywood don't even use raw anymore? If professional Hollywood colorists are ok with ProRes why is it the indies seem to think raw is a must to achieve success?

The Pocket is going to be a great addition to the m43 camera world. I'm very excited to have the option of a raw camera for $1,300. Its kind of a no brainer if one actually needs that sort of thing. My whole point from the start of this thread has been how many people actually need it? I mean really need it and not just desire it. There is a significant difference there. A lot of people buying the Pocket will likely just fall in the desire it because they can have it category without really realizing what that actually means. I see a lot of BMD stuff out there from people that would probably be better off using a camera that did not require a lot of manual grading. Luts do help a lot but if one just plans on slapping on a lut how is raw providing them any real advantage over a 10bit log format?

So to say no compressed codec can compare to raw is a bit misleading because it depends on what one needs to do with that video. If Hollywood movies that actually earn money at the box office are ok with ProRes and if the GH5 10bit formats hold up equally well to ProRes then the GH5 can be an acceptable camera and format to shoot with for film making. At least Cinegal and Stephen Fleet think so. If anybody has any tests to prove otherwise I would love to see them. I'm not talking extreme cases where the grade is beat to death to break the codec but actual real world use where a well exposed and white balanced shot is graded for a look.

 

[Sage]

Just sent these out to everybody, with the V2 update;

Here are the raw footage side-by-sides:

Daylight Alexa & GH5 Raw Footage
Tungsten Alexa & GH5 Raw Footage
(Use DaVinci Resolve for these - accurate Arri ProRes)

 

[John Brawley]

The Ursa Mini 4.6 experience wasn't exaggerated,

I don’t know Tom, you list some pretty big points. I thought I’d look for what you’re referencing.

game over for Arri and Red. Where were you? I lived it for months on end. Philip Bloom wrote about it,

What exactly did he write ? I saw he did a review of the 4.6 (which I dind’t watch) and at the very top of the review he also stated that...

“EDIT: THIS VIDEO REVIEW WAS DONE WITH THE FIRMWARE SUPPLIED WITH THE RENTAL CAMERA. THERE IS NEWER FIRMWARE COMING OUT SOON WHICH LOOKS LIKE IT IMPROVES ON MANY OF MY OPERATIONAL ISSUES WITH THE CAMERA.”

http://philipbloom.net/blog/ursa/

In fact he seems to end up with a more positive experience than a negative one ?

“My thoughts and findings are in the video below but to sum it up…this is easily the best camera they have made. It is very well priced and produces a wonderful image. It is far from perfect, and I go into the pros and cons in the video. Most of the cons are operational rather than image…mostly!”

the FPN that could not be corrected in post,

Yep. True. You can’t fix FPN easily in post.

Mostly when I’ve seen FPN in BMD users work it’s because their blacks aren’t black. You can easily see FPN in a BMD camera by lifting the blacks up. I won’t deny at all it hasn’t been a bigger problem on previous cameras from BMD, but I’ve now used many bodies for a couple of years and it just doens’t come up once you grade it.

I’ve just finished shooting my umpteenth series with them and I can once again post plenty of stills in low light but I think everyone’s sick of hearing from me by now.

the magenta hue.

Yes that was a big one. It wasn’t a hue, it was that the corners went magenta in some shooting circumstances. Mostly because of Sensor / Lens shading. Telecentric lenses minimises it. They themselves have done a lot to minimise it since it was brought to their attention by users. Again, it’s never been an issue for me in my work but from memory 35 - 50mm at F8 was what brings it out most.

Repeated delays, broken promise dates. The global shutter feature that caused great excitement; rescinded.

I think these were related. They were trying to make an announced feature work and it was going to take too long. Instead of holding up the release of the camera by many months (maybe more) they removed that feature before shipping.

15 stops of DR that didn't hold up under EBU testing.

I read the report. He actually said “more than 14 stops”. In the how long is a piece of string world of DR tests I think that’s within subjective tolerance. There’s no accurate test for DR because it’s a highly subjective choice about what noise floor you accept. This is why we get so many different opinions about DR and other silly phrases like useable DR.

Remember when RED were claiming 20 stops form their old Dragon Sensor ?
http://www.reduser.net/forum/showthread.php?92158-Enter-the-Dragon

I also seem to recall when Alan talked about lurid colours he was testing with VIDEO and not FILM (Log) because it’s for the EBU. That was the inference anyway. The B in that stands for Broadcast, and that’s the emphasis they take in testing, on suitability in a broadcast environment. He mostly seemed to be testing in VIDEO mode and interested in how that performed, not in FILM mode.

Let’s look at REDUSER and you see these kinds of current posts.

This one from their newish premium camera the Helium...
http://www.reduser.net/forum/showthread.php?163665-Helium-Sensor-Smearing

Or

One of the many options some choose instead of an Ursa Mini, the Scarlet W reporting similar issues from multiple users.
http://www.reduser.net/forum/showth...d)-Scarlet-W-Experience&p=1800200#post1800200

It’s not hard to find these examples with many manufacturers. I can rattle of some bad Sony F55 RAW and Arri experiences if you like.

I don’t think anyone is denying that BMD have had more than their fair share of issues, but I think they’ve also made great leaps in confronting and dealing with those issues as well.

JB

 

[Tom Roper]

The emperor was wearing new clothes when he stated the GH5/s are "hobbled by a heavily compressed consumer codec." 12 bit raw is not game over for 10 bit 422 log 400/mbps all-I. And there is *not* every reason to expect that BMD will get this right. There is every reason to be circumspect, while hoping that they do. We will know when it starts shipping on time, if it ships on time. It could indicate trouble just as it did for the Ursa Mini 4.6k if it does not. Game over this is not, players not even on the field.

 

[joe12south]

The emperor was wearing new clothes when he stated the GH5/s are "hobbled by a heavily compressed consumer codec."

I have no idea what that means. Are you referencing "the emperor has *NO* clothes"? Even if you are, the statement makes little sense.

12 bit raw is not game over for 10 bit 422 log 400/mbps all-I. And there is *not* every reason to expect that BMD will get this right. There is every reason to be circumspect, while hoping that they do. We will know when it starts shipping on time, if it ships on time. It could indicate trouble just as it did for the Ursa Mini 4.6k if it does not. Game over this is not, players not even on the field.

You're conflating my statements, with the OP's title choice. Let's unpack, shall we?

1. The GH5(s) internal recording has challenges compared to RAW (even compared to ProRes, but less so there.)
A. 400mbps is not a lot of bandwidth for 4K ALL-I. In fact, other than motion rendering, it can sometimes looking worse, with *MORE* compression artifacts than the Long-GOP version.
B. Macroblocking. Low contrast areas, shadow areas, both areas an aggressive H264 codec is going to compress the hell out of. Add motion to the mix, and you've got blocking you're not going to fix in post. (You can try to crush it out of notice.)
C. Heavy load on editing processor. Even if your computer is fast enough to edit it in real time, that's processing power that could better be used for other things (like more real time filters.)
D. When you've got to push LOG as far as you do to get a usable image (especially V-LOG L which intentionally gives up almost 30% of bandwidth), there is a meaningful difference between 10-bit LOG and 12-bit RAW. Probably not if you're just making minor tweaks and applying a final grade LUT, but if you need to do serious manipulation and recovery (more and more common these days) then the RAW advantage is real. We all see it and know it from stills processing, I don't understand why so many can't understand how that same advantage can apply to moving images.

2. So the "game over" title was click-bait. We all know that. Move on. Have a meaningful discussion. That said, with all that the BMPCC 4K will offer, for less than half the cost of the GH5S, BM has at least *changed* the game. Panny chose to gouge the "filmmaker" market with the GH5S pricing, and chose to *not* give it a professional codec. BM chose a different strategy. If a cheap camera is able to record an uncompromised signal directly to a fast/cheap HD, *MANY* filmmakers are going to choose it over the GH5S.

3. These are cameras, folks. Tools. Not our children. We can pick the best tool for a given job, and not worry about feelings. We don't owe corporations loyalty. The GH line has done a lot right for a long time. Hard to imagine that I won't own a GH6. BUT, the BMPCC 4K has the features many want at a savings that is hard to ignore. It won't take many GH5 sales, but unless something is seriously wrong with the production product, it's going to steal a lot of GH5S sales.

 

[combatentropy]

I have no idea what that means. Are you referencing "the emperor has *NO* clothes"?

https://en.wikipedia.org/wiki/The_Emperor's_New_Clothes

400mbps is not a lot of bandwidth for 4K ALL-I.

Panasonic said that 100 Mbps AVC-I was a high-quality codec for its professional news cameras and that 200 Mbps was visually lossless, at HD resolutions. Therefore I would expect 400 Mbps 4K to be decent. We may quibble over where to draw the line, but it doesn't make sense for Panasonic to be so far off that they call it "high quality," and you call it "not a lot" and "aggressive" compression. --- https://pro-av.panasonic.net/en/sales_o/news_info/news2014/20140228_03.html

It's a ratio of about 10:1. Even JPEG is said to have "little perceptible loss" at that ratio, https://en.wikipedia.org/wiki/JPEG

there is a meaningful difference between 10-bit LOG and 12-bit RAW

Yes, and 10-bit log should actually hold up better than 12-bit linear, http://www.normankoren.com/digital_tonality.html#Human_vision, so important is logarithmic over linear.

 

[Cary Knoop]

Panasonic said that 100 Mbps AVC-I was a high-quality codec for its professional news cameras and that 200 Mbps was visually lossless, at HD resolutions. Therefore I would expect 400 Mbps 4K to be decent. We may quibble over where to draw the line, but it doesn't make sense for Panasonic to be so far off that they call it "high quality," and you call it "not a lot" and "aggressive compression." --- https://pro-av.panasonic.net/en/sales_o/news_info/news2014/20140228_03.html

Of course it is high quality for viewing but not necessarily high quality for editing.

 

[Thomas Smet]

I say prove it. I posted videos that show how robust the quality of the codec on the GH5(s) is. Instead of just saying it isn’t good enough I think people should post examples of how it isn’t good enough.

Again I’m not saying it is as good but that I think we really need to put some visuals behind just how it looks compared to other formats instead of just using the same old talking points.

I posted some videos of some professionals showing how good the Gh5 codecs actually are. Now it’s time for those who find fault with the codecs to show us just how much the formats suffer.

 

[wwjd]

I got a GH4 >>BECAUSE<< Blackmagic Design fumbled the ball on their 4K. Then, it was riddled with problems for years. I know they make some great things and a higher percentage of them actually work, but I'm waiting for the reality of release. GH4 still works fantastically with it's DR of 10.

 

[Barry_Green]

A. 400mbps is not a lot of bandwidth for 4K ALL-I. In fact, other than motion rendering, it can sometimes looking worse, with *MORE* compression artifacts than the Long-GOP version.

Just a point of fact here -- Netflix is one of the biggest studios around, spending $8 billion on production this year alone. They have conducted extensive testing on the EVA1, and had been evaluating the 400mbps codec internally for qutie a while. When Panasonic announced the 400mbps ALL-I codec publicly, Netflix issued a press release that day saying that the EVA1 is now a Tier 1 camera, approved for all Netflix-funded productions, when using that codec.

I don't know if the GH5's 400mbps ALL-I is the same as the EVA1's, but I suspect that it is very likely identical. If so, the GH5's codec has undergone rigorous testing and has passed.

For Panasonic's part, they issued a white paper comparing the 400mbps EVA codec favorably to 320mbps AVC-Ultra.

It may not be perfect, but it certainly isn't bad by any means.

 

[Tom Roper]

All-I 400 mbps 4:2:2 qualifies for UHD1 Tier 1 for EBU 3335 at up to 25 fps.

 

[Greg_E]

Still arguing about how the BMD beats everything, and yet we still haven't seen any good footage that users can download and tear apart. What if the image is loaded with artifacts and moire (redundant)? What if it is so full of IR pollution that you might as well call it an IR camera? I'm waiting to see what it looks like before passing any judgement about how it may or may not be the best thing since individual coffee packets and machines to dispense them. (sliced bread is very old by now, need a new phrase)

I think many (most?) of us users are genuinely hoping they (BMD) did everything as right as the form allows and it produces beautiful images in compressed and raw workflows under every varying condition that could be called decent lighting, and at least decent sound from the single XLR. We are hoping this really is the game changer camera of wicked high quality and very low price. But I think we may need to wait and see, and may need to wait for that first firmware upgrade after release (or maybe even longer).

 

[DeeZiD]

In my tests both the 150mbit and 400mbit looked identical in almost 99% of the shots...
So I really don't see how the 400mbit codec is inferior.

For the GH5(s) I just wish for a bypass option of sharpening and temporal nr.

 

[Vardalos]

In my tests both the 150mbit and 400mbit looked identical in almost 99% of the shots...
So I really don't see how the 400mbit codec is inferior.

For the GH5(s) I just wish for a bypass option of sharpening and temporal nr.

This test by Steven Fleet compares 150Mbps to 400Mbps for keying on the GH5. https://youtu.be/uXhn-5UILgk BTW, Steven Fleet; is a VFX Supervisor on TV series like Marvels Iron Fist on Netflix. This guy is a serious Pixel Peeper, his job calls for it.

 

[John Brawley]

Once again,

There’s a difference between your eyeballs seeing a difference and what you can usefully do in post with those same files.

Bit depth matters for colour. It matters a lot, especially for high DR images. As revolutionary as 10 bit is for those of you experiencing in it for the first time with the GH5, 12 bit log is the SAME step up again.

While I’m sure there are great ways to measure how good the 400 Mbs codec is and I’m sure it’s good, you can’t argue with bit depth, independent of how compressed it is.

For some people, that will matter. For some people it won’t. For some it’s a massive win. For some it will be a massive workflow headache.

I recall in the early days of the 5D and GH hacks, those talking about how good an 8 bit Log-like profile could be and how well it could be graded.

Yes, it can look great and be great no doubt to your eyeballs. But you run out of wriggle room very quickly if you have a challenging scene to grade. 10 bit helps a lot. 12 bit helps a lot more.

JB

 

[TheDingo]

But you run out of wriggle room very quickly if you have a challenging scene to grade. 10 bit helps a lot. 12 bit helps a lot more.

For me it's always "the proof is in the pudding", where you compare two different cameras shooting the same scene under identical conditions, and then see if reality backs up how things appear on paper. I have seen some surprises where higher spec'd gear was not able to outperform the lower spec'd gear, and much of this had to do with the actual implementation of those specs in the finished design.

I plan on picking up a BMPCC 4K body to go with my GH5 camera kit, but how and when I will use it will largely depend on how it compares with the GH5/GH5s. Personally I think all bets are off (with the exception of RAW recording) until we can compare actual footage from both cameras.

 

[Thomas Smet]

Once again,

There’s a difference between your eyeballs seeing a difference and what you can usefully do in post with those same files.

Bit depth matters for colour. It matters a lot, especially for high DR images. As revolutionary as 10 bit is for those of you experiencing in it for the first time with the GH5, 12 bit log is the SAME step up again.

While I’m sure there are great ways to measure how good the 400 Mbs codec is and I’m sure it’s good, you can’t argue with bit depth, independent of how compressed it is.

For some people, that will matter. For some people it won’t. For some it’s a massive win. For some it will be a massive workflow headache.

I recall in the early days of the 5D and GH hacks, those talking about how good an 8 bit Log-like profile could be and how well it could be graded.

Yes, it can look great and be great no doubt to your eyeballs. But you run out of wriggle room very quickly if you have a challenging scene to grade. 10 bit helps a lot. 12 bit helps a lot more.

JB

Proof?

 

[John Brawley]

Proof?

I’m not sure what you’re asking here Thomas ?

JB

 

[joe12south]

The responses here baffle me. Do this:
Take your GH-whatever. Shoot a still recording, under exposed 2 stops, as both a JPEG and a RAW image. Lift that image in the image editing app of your choice (hell, your NLE if you want.) Tell me which one you'd rather work with.