Heya, i'm the weird beardy guy in the hillbilly grip truck video. The recompression and youtube we used didn't help the footage, but going back over the source, the 12,800 looked quite serviceable with a small amount of work. To my eye, it was better straight from the camera than the gh4 is at 800, even if you denoise in post. That puts its closer to 4 stops better performance... or i'm just not in tuned with the gh5 shortcomings the way I am the gh4.
GH4 Panasonic GH5 coming in 2017
- Thread starter Dan Keeble
- Start date
philiplipetz
Veteran
THANK YOU for one of the most informative GH5 videos. BTW your bearded face is the ideal test pattern. Can you give us any more technical information on what you used during shooting and post?
Last edited:
Mark Williams
Veteran
Nice test, that helped me a lot in understanding the GH5.
filmguy123
Veteran
assuming you are doing post work, would there ever be an advisable time to use the in camera electronic stabilization instead of or in addition to post electronic stabilization (ie warp stabilizer)?
joe12south
Veteran
Definitely, if only for one big reason: compression artifacts.
The more motion between frames, the higher the data rate, the more noticeable the compression artifacts. In camera electronic stabilization sits before compression in the image processing pipeline, so you stop that loss from ever happening.
joe12south
Veteran
A lot of folks would LOVE you...or at least buy you a beer...if you'd post lightly compressed (ProRes) snippets from your test. Vimeo allows the source files, or you could use some other free big file hosting service.
Thomas Smet
Veteran
Another issue in post is motion blur. When that camera is shaking around you are getting motion blur on each frame. Stabilizing in post can help keep things steady but you are not getting rid of the motion blur so each frame will slightly shift in and out of softness. The worse the camera was moving the more random blur you will get for each frame.
Also if you have to do extreme stabilization you may end up cropping into the frame slightly to compensate for the blank areas on the edges. This can alter your intended FOV of each shot depending on how shaky you were.
Stabilization in camera is always, always, always a better solution. It means no random motion blur, no unpredictable cropping and no resolution loss by scaling up and cropping. Post stabilization is meant to be a fix as an exception to the rule and not meant to be a normal workflow.
Rambly reaction
Rambly reaction
Cheers, guys.
I didn't bother keeping any of the files, as it was just a silly test, but we did it at my mate's house so i'll check if he kept any. Btw, did you see in the video description the link to the master mp4 file on google drive? I reckon I could see as much from that as I needed to know. I mean, its not like this is great footage to grade and there is something clearly up with the 10 bit on the preproduction camera so you can only garner so much from the test. But, getting away from youtube recompression makes a heap of difference, so you might wanna check the mp4 out.
12,800 @ 2.8 is mighty sensitive. we shut off the lights and pulled the curtains until it was almost black, then used 2 z96s to light. By the time we hit 6400 we'd dimmed the leds to minimum and had to move them across the room AND up the shutter to keep the 12,800 from blowing out. crazy.
I don't doubt we could have tweaked around and found slightly better settings for both cameras, but we didn't have much time left and had been shooting up in the mountains all day so the goal here was "set both cameras to as factory stock as possible with all the same settings and see if the gh5 is improved in low light". I didn't want to have our post process influence the test, so we shot on the "standard" photo profile instead of one of the cine profiles. That way, the post was just "put the clips on the timeline, cut out some of the boring stuff i said, add some CG overlays and export." Rowan didn't want to bother with the 10 bit since in theory it would be identical without a grade but I thought it might have some different curve or something that might give it a wider dynamic range in post. He was right - they look identical except for the noise difference, which remains a mystery, but isn't something i'll give much of a thought until the final camera is released.
If there are any specific questions i can answer, i'm happy to. There isn't a whole lot more to say though. I was doin this test for myself and my mates as much as anyone (as the hits on the blog/channel aren't great and aren't monetized anyways), so i was damned careful to set up as apples-to-apples as possible. Anyways, happy i could contribute to the conversation. You can only pixel peep a preproduction camera with a grain of salt, but I still reckon the improvements across the board are significant... especially the low light. Cheers!
Rambly reaction
Cheers, guys.
I didn't bother keeping any of the files, as it was just a silly test, but we did it at my mate's house so i'll check if he kept any. Btw, did you see in the video description the link to the master mp4 file on google drive? I reckon I could see as much from that as I needed to know. I mean, its not like this is great footage to grade and there is something clearly up with the 10 bit on the preproduction camera so you can only garner so much from the test. But, getting away from youtube recompression makes a heap of difference, so you might wanna check the mp4 out.
12,800 @ 2.8 is mighty sensitive. we shut off the lights and pulled the curtains until it was almost black, then used 2 z96s to light. By the time we hit 6400 we'd dimmed the leds to minimum and had to move them across the room AND up the shutter to keep the 12,800 from blowing out. crazy.
I don't doubt we could have tweaked around and found slightly better settings for both cameras, but we didn't have much time left and had been shooting up in the mountains all day so the goal here was "set both cameras to as factory stock as possible with all the same settings and see if the gh5 is improved in low light". I didn't want to have our post process influence the test, so we shot on the "standard" photo profile instead of one of the cine profiles. That way, the post was just "put the clips on the timeline, cut out some of the boring stuff i said, add some CG overlays and export." Rowan didn't want to bother with the 10 bit since in theory it would be identical without a grade but I thought it might have some different curve or something that might give it a wider dynamic range in post. He was right - they look identical except for the noise difference, which remains a mystery, but isn't something i'll give much of a thought until the final camera is released.
If there are any specific questions i can answer, i'm happy to. There isn't a whole lot more to say though. I was doin this test for myself and my mates as much as anyone (as the hits on the blog/channel aren't great and aren't monetized anyways), so i was damned careful to set up as apples-to-apples as possible. Anyways, happy i could contribute to the conversation. You can only pixel peep a preproduction camera with a grain of salt, but I still reckon the improvements across the board are significant... especially the low light. Cheers!
joe12south
Veteran
Unfortunately, the in-cam electronic stabilization will not reduce motion blur. The IBIS stabilization will, of course.
joe12south
Veteran
I missed that. Thanks!
joe12south
Veteran
Camera induced motion blur. Subject blur is a result of shutter speed.
filmguy123
Veteran
Oh wow, answer surprised me. So YES to in camera electronic stabilization?
Will this not introduce weird artifacts in the recording?
Will this not introduce weird artifacts in the recording?
fboonzaier
Well-known member
Seems like the video was intended as a joke?...not sure why Panny would give him a GH5...maybe a Millenials thing?....I say that because my 19 year old daughter would probably enjoy his "review"...
Steve Ogden
Well-known member
Joseph Linaschke just posted some AutoFocus testing at UHD60p
He said he shot it before he really understood the settings. The video where Panasonic explained the settings is https://youtu.be/6PyqWrztPkc
He said he shot it before he really understood the settings. The video where Panasonic explained the settings is https://youtu.be/6PyqWrztPkc
AndreeOnline
Veteran
That's a great way to produce content quickly. And that's all that's great about it.
Too bad then, that this is the way 95% of content is created. The upside being that 95% of people watching those videos don't understand that the person speaking also doesn't understand. Perfect symbiosis.
Steve Ogden
Well-known member
Well, in his defense he had access to a pre-production camera that came without a manual or any explanation about how the settings work. He didn't represent it as anything it wasn't, and fully disclosed that he's making it available and sharing his learning as he goes. I see his admission of not initially understanding the settings as intellectual honesty and very different than posting from ignorance. I never hold it against anyone that they don't know something before they know it and certainly respect the willingness to share as they learn.
Last edited:
The overlay on the video also pointed out that the settings were the exact opposite of what they should have been. I think most all of these early videos should be taken as rough tests both from the early firmware and lack of knowledge of users of how the settings work ( if they work at all with the early firmware). I have one on order and look forward to learning how to use which I am sure will take some time. I know it took me best part of a year to get the most out of my Sony NX5U so don't expect to get it wonderful the first few attempts.
Peter Berry
Well-known member
+1
And if anywhere near repeatable, show an amazing 8-bit hi-ISO IQ gain over the GH4...
AndreeOnline
Veteran
I completely agree with this.
But I do think it's of significant importance, going trough life, that you learn to recognise how to separate what you know from what you don't know. I haven't seen the linked video and based on your answer, it's quite possible that this person belongs more in the 5% group than the 95% one. I simply went with the odds—that oddly enough seem to remain constant—and weighed heavily in my favour.
-----
Just stopped myself from going off on an unimportant tangent lined with anecdotes. Let's instead focus on the camera: any news on why the 10 bit files looks worse than the 8 bit ones (seen in another test)?