GH2 For Color Grading: Hacked GH2 or Canon DSLRs...?

Ryan-Guy said:
I have a 5d & 7d and grade it on Apple color and I don't think i could ever stretch my images like this.
Click to expand...

Yes you can. Sure you may get some macroblocking if you use just one channel but proper color correction tools can work wonders. You could easily stretch quite a lot of 8-bit cameras that much and even way further. Just take a look at how Stu color corrects some 5d stuff on an older version of Colorista. Colorista II also works considerably better on DSLR material than Apples dated Color.

http://www.redgiantsoftware.com/videos/redgianttv/item/23/

Pushing colors here and there isn't the tricky part. Where 8-bit compressed cameras tend to really break apart is making masks from color channels etc. It's more about knowing what you are doing (knowing the tools and what to do) than the camera.
 
I am going to be the other naysayer in this bunch. I have now graded GH2 and 5D in Resolve, Baselight and just about everything else I could get my hands on. In the hands of a proper technician the 5D image will still hold up better. We are not talking throwing something into Magic Bullet here. The reason I say this is because the GH2 has noise and banding issues. As soon as you really begin to push around the curves it shows up. Which means that you get colored noise as the image falls apart. The red pillar in the image a couple of pages back will show you this if you look closely at the jpeg. (http://img220.imageshack.us/img220/6470/yellowafter.jpg) There is light magenta noise beginning to appear everywhere in the red pillar for instance. You may not see it here on the net, but I guarantee you will see it on a big screen. The same is even more true for gradients. You will get complementary color crossover if you push any grey surface with a gradient using this camera. Hack or no hack. Further, the shadows of the vest in this image flatten out and take on a green hue as the image is optimized for skin tones (magenta into the face). Traditionally, the green channel is the one hardest hit by 4.2.0. So the shadows end up having very little gradation and become flat patches if the sensor cannot provide them with enough information.

These are the limits of the size of the sensor, 8 bit 4.2.0 and a non-linearized profile.

The 5D has other issues, but as soon as you pop in the Mosaic filter some of them are gone. Even with all the line skipping, macroblocking and in camera noise reduction trickery the solidity of the image due to it's larger sensor, will produce a tighter image to grade from and therefore the appearance of greater effective pixel density, hence smoother transitions. What most people are responding emotively to in the GH2 is the localzed contrast and the sharpness of the image. This is certainly enhanced by the hack as it raises detail with the higher bit rate.

Of course I am aware that some people will be displeased by my comments, but this has been my experience and I don't really feel like entering into a full scale debate. So yes I have tried quite a few different patches on the GH2 as already previously stated, and my conclusion is that if you know how to properly light and expose with both of these cameras the 5D still holds up better. Which doesn't mean I don't like my lovely little GH2 as well.
 
Last edited:
yoclay said:
I have now graded GH2 and 5D in Resolve, Baselight and just about everything else I could get my hands on.
Click to expand...

Have you graded hacked footage?

We are not talking throwing something into Magic Bullet here.
Click to expand...

Are you trying to be insulting?

The reason I say this is because the GH2 has noise and banding issues.
Click to expand...

What was the ISO that the GH2 was shot at? Any that are affected by the ISO bug? How competent was the shooter with the GH2? Was the camera hacked? If so, which one?

The red pillar in the image a couple of pages back will show you this if you look closely at the jpeg. (http://img220.imageshack.us/img220/6470/yellowafter.jpg)
Click to expand...

Why are you using a JPEG sample for critical analysis? Are you unaware of what JPEG compression can do to an image that may not be in the original?

The 5D has other issues, but as soon as you pop in the Mosaic filter some of them are gone. Even with all the line skipping, macroblocking and in camera noise reduction trickery the solidity of the image due to it's larger sensor, will produce a tighter image to grade from and therefore greater effective pixel density, hence smoother transitions.
Click to expand...

The GH2 has a greater video resolution than the 5D. It doesn't matter how large the 5D sensor size is, at the end of the day it's around 700 lines of resolution with made-up information getting it to the 1080 of the video file. Are you not aware of this?
 
yoclay said:
I am going to be the other naysayer in this bunch. I have now graded GH2 and 5D in Resolve, Baselight and just about everything else I could get my hands on. In the hands of a proper technician the 5D image will still hold up better.
Click to expand...

I've been through DaVinci on both, out from a 2K projector in a theater... hacked footage vs 5D footage is no contest WHAT so ever.

Maybe NON hacked, but I completely disagree with you and I'm sure any serious technician would tell you that isn't the case.
 
I've got both, shot both. I'm afraid yoclay's conclusions are incorrect. Unless you are talking about unhacked camera.
 
That is my experience of working with both and my conclusion for the time being. Everyone is free to have their own differing viewpoint. Sorry not going to join the debating society today.
As I have already stated for a third time, yes I am speaking about a hacked GH2 with a variety of different patches. And no I do not consider working with Magic Bullet anywhere near what can be achieved with Resolve, Baselight or Lustre.
 
yoclay said:
And no I do not consider working with Magic Bullet anywhere near what can be achieved with Resolve, Baselight or Lustre.
Click to expand...

I guess you missed the post where Kholi mentioned using Resolve.

You've seen plenty of examples from the "pro" end. Care to post those of your own? No debate required, just post PNGs of your work and let it speak for itself.
 
yoclay said:
And no I do not consider working with Magic Bullet anywhere near what can be achieved with Resolve, Baselight or Lustre.
Click to expand...

You don't have to debate, that's fine, but what you're saying is not true. Even yoru example using the Mosaic filter still does not have baring on how the data lands to recording media. It JUST gives you a SLIGHTLY better base form which to work, it doesn't change the fact that it's still being crammed into a 4:2:0 low-bitrate format.

And I don't work in LITE either, talking about full suites. You get better results using better software working in higher floating point spaces with proper compression (DPX preferably, and really is the only way to roll for finishing IMO)... regardless, it doesn't matter if it's MB or AE or anything like that, the results will scale equally.

If GH2 hacked footage is a healthier base than 5D footage for 8-bit space, the same results scale.
 
I owned both and graded both and canon is far better in my experiance for grading. The gh2 pre hacked breaks down miserably with the first pass of color correction. Not sure about the hacked as I was never able to get it going.
 
reem12 said:
I owned both and graded both and canon is far better in my experiance for grading. The gh2 pre hacked breaks down miserably with the first pass of color correction. Not sure about the hacked as I was never able to get it going.
Click to expand...

This thread is labeled "hacked Gh2", not pre-hacked... I'm just saying.
 
Actually, what would be a proper 5D vs GH2-Hack grading test? One that's VERY simple to do?

I don't have color charts, but I have both cameras and I can use the same glass (new EF primes) on both.

Will purposely use a close but improper white balance on both and light with my Coollights HMI, which has a serious green shift, and an E47 socket bulb. Just a house lamp. Idea is to get it from wrong to cinema right?

Squash this noise right now.
 
It would be great if someone could get a screengrab using a Canon, and then with the same lens, get a screengrab using the hacked GH2 (same framing, same image) .. and post the .tif files... I'd surely like to do a test... And obviously, each camera should use a picture profile optimized for grading...
 
Last edited:
Breezer said:
It would be great if someone could get a screengrab using a Canon, and then with the same lens, get a screengrab using the GH2 (same framing, same image) .. and post the .tif files... I'd surely like to do a test... And obviously, each camera should use a picture profile optimized for grading...
Click to expand...

I've thought about doing this since I have a 60D and may possibly be able to get a hold of a 5D to use as a test.
 
Kholi said:
Actually, what would be a proper 5D vs GH2-Hack grading test? One that's VERY simple to do?

I don't have color charts, but I have both cameras and I can use the same glass (new EF primes) on both.

Will purposely use a close but improper white balance on both and light with my Coollights HMI, which has a serious green shift, and an E47 socket bulb. Just a house lamp. Idea is to get it from wrong to cinema right?

Squash this noise right now.
Click to expand...

Just posted that. Soon as I get a second here I'll do it.
 
For myself, I was thinking at least 60D and GH2, with Marvels Cine (latest version) and MyFilm1 set all the way down, and with a Nikkor 50mm f/1.2 lens for both.
 
Good question: which profile do you guys want from the 5D? I think Cinestyle is the worse but so many people swear by it. Let me know which profile and I'll use it.
 
The 5D/7D are great... but they don't even touch on the gradblity of the hacked GH2.

Flat:


Graded:


Grading you would never use, but still holds up:
 
Last edited:
Kholi said:
Good question: which profile do you guys want from the 5D? I think Cinestyle is the worse but so many people swear by it. Let me know which profile and I'll use it.
Click to expand...

Advance Marvel Cine please

Gary Huff said:
For myself, I was thinking at least 60D and GH2, with Marvels Cine (latest version) and MyFilm1 set all the way down, and with a Nikkor 50mm f/1.2 lens for both.
Click to expand...

That'd work !
 
Kholi said:
Just posted that. Soon as I get a second here I'll do it.
Click to expand...


Whatever color profile you want, I personally use the most popular one I guess, neutral, sharpness -4, contrast -4, sat 0 or -1, color tone 0 ... I never tested the others...

I'd say try Low ISO, well lit scene, just in case the 5D is a better performer in low light... Faces/skin tones would be good to test when color grading but anything will do... Just a nice gradable image ;)
 
You must log in or register to reply here.
Back
Top