GH2 Are all these new hacks and patches creating a placebo effect?

[omnidecay]

I think the real issue comes from people NEEDING some type of comparison. Why do you need to compare hacked footage vs unhacked? When your looking to buy a new car do you compare Lamborghini's vs pickup trucks? Its not a one is better then the other type of thing. It all comes down to which one is more practical in your specific situation. Hacking is great if you can afford the time for it but unhacked can work just as well depending on what you can reinvest your saved time into. Getting the project finished is the number 1 goal of any filmmaker the rest is all risk vs reward.


"A movie is never finished, only abandoned"

- George Lucas

 

[Ralph B]

Isn't the HDMI output limited to 1080i? It couldn't be an "an absolute frame of reference" if that is true.

You have a misunderstanding of the HDMI output. It's true 24P wrapped in a 60i stream (or 50i stream for PAL users).

 

[mpgxsvcd]

+1

Bout time the voice of reason showed up.

People, you are not going to see any difference especially if all you do is view these images on a small computer screen. If you are not shooting for a 40ft screen, broadcast of just large TV screens then you probably are wasting your time by hacking your camera. Those reasons are why a Red, Alexa and other expensive cameras exist. But the fact that we can eck out "similar" (not necessarily the same) kind of quality out of this $700.00 camera is a bonus. If you don't see the difference then more power to you. As the Dingo suggests....there is a point of diminishing returns. You will not see those small differences on a computer screen. So, as is, the camera might be "good enough" for you.

If I get close enough to my 1920x1080 monitor then I can tell more from my computer screen than I could from the largest big screen out there. It is all about matching the display to the source resolution of the footage and getting as close to the display as focusing will allow. You don’t need a big screen to examine the footage. Your monitor will do just fine as long as it is 1920x1080 for 1080p footage and 1280x720 for 720p footage.

 

[mpgxsvcd]

You have a misunderstanding of the HDMI output. It's true 24P wrapped in a 60i stream (or 50i stream for PAL users).

No I understand that. The problem is that the footage has been altered from its captured form. It is not the 1080p @ 24 FPS that the sensor recorded. Now that output could be lossless to the original or it may not be. Only testing would determine that.

 

[mpgxsvcd]

I think the real issue comes from people NEEDING some type of comparison. Why do you need to compare hacked footage vs unhacked? When your looking to buy a new car do you compare Lamborghini's vs pickup trucks? Its not a one is better then the other type of thing. It all comes down to which one is more practical in your specific situation. Hacking is great if you can afford the time for it but unhacked can work just as well depending on what you can reinvest your saved time into. Getting the project finished is the number 1 goal of any filmmaker the rest is all risk vs reward.

No I shop all the cars in my price range. I might look at a Honda and determine that it is good value for the money but it simply doesn’t go fast enough for my tastes.

I may look at a BMW and determine that it has all of the performance I desire but I just don’t want to pay that much for a car.

Then I might look at an Infiniti and determine that it has 96% of the performance for 75% of the price of the BMW and decide that it is the car for me.

Other people might buy the BMW strictly because someone said it is the “Ultimate Driving Experience”. While other people might decide to get the Honda because they really don’t think they need any more performance than it offers and they don’t want to waste money on “Diminishing Returns”.

 

[Lpowell]

The truth is that the Color, Resolution, and Dynamic Range claims can be tested for entirely static scenes. I completed the data gathering phase of those tests last night. I will be uploading the files as they came out of the camera over the next 1-2 days(The darn high bit rate files take forever to upload).

Then I will create my video analysis of what I think of the samples for anyone who is interested in what I have to say about it. For everyone else I will give you the tools to accurately judge Color, Resolution, and Dynamic Range for yourself.

Please be patient with me getting the data up there. It took a long time to shoot it all and it takes even longer to edit the conclusion video and post these gigantic files. Trust me in the end you will want to see these tests. I believe they will conclusively show that there is a difference between hacked and not hacked footage. However, it will also show that some of the claims that have been made are completely unsubstantiated and that others are indeed valid.

Thank you in advance for your systematic work, mpgxsvcd, this is the kind of testing that provides genuinely useful benchmarks for making objective judgements. As other well-publicized shootouts have shown, contemporary video cameras have unexpected strengths and weaknesses. It's always advantageous to understand the limits of your tools in order to make informed choices in how you use them.

In addition to increasing encoder bitrate, PTool provides a wide variety of ways to fine tune the performance of the GH2. Compared to the GH1, whose issues primarily involved maximizing both bitrate and stability, the GH2 presents a far more complex optimization challenge. With the recent release of Panasonic's v1.1 firmware, supporting 1080p25 and 1080p30 AVCHD modes, we're starting a new phase of patch development which will require another round of careful testing before producing conclusive results. The clearly visible evidence we've seen of consistently measurable improvements in quality is an impressive achievement for this early stage of development.

 

[Kholi]

Who here had an HVX200, HVX200A, and HPX170?

I had all three cameras. I used all three extensively. Of each, hands down, I swore up and down that the 200A looked better than the HPX170. Something happened in between the 200A and 170 that wasn't supposed to, it was noise and a color shift, I can't remember exactly what but they were supposed to be the exact same image. A ton of people didn't notice any difference at all, but there was just enough of a subtle change that it drove me mad.

I sold my HPX170 faster than I've ever sold any camera before. My HVX200A's image was great, but it only took that inkling of a difference to change my mind on the 170. To me, the Hack is the same way. There's enough of a change in the overall IQ, which could all be attributed to less compression and seeing more of the camera's true nature, to make me want to use it instead of tools that are obviously technically much much better. It may not be rocket science, but there's nothing that could've shown me different had I not picked the camera up before and after and looked at it myself.

The claims of increased DR, color accuracy, so on and so forth aren't what I'm looking at. It's strictly the fidelity of a captured AVCHD stream, which mainly comes down to the encoder's ability--when hacked, to retain detail as the camera shifts, or anything in the image moves around. It's also much MUCH more robust in lowlit situations, where lowlit's defined to me as having only five thousand watts of light, and I've got to go and light a large living room space with that.

Going back to old footage from my feature, the interior footage from the GH2 is outright atrocious. My first hand experiences in production environments are enough for me to be able to judge a camera's image quality, and compare it to another. I don't need double-blinds to know that 5K footage rips 4K footage, or 4K footage down to 1080 looks much more resolute than current crops of 1080 cameras (excluding the C300 at this moment).

I understand that some people do, and some people can't immediately see the difference in footage...

Well... I was working on a project that was shot with the Canon 7d. Edited it and was projecting it around when our second editor came in the room. Looked at the image on the big screen and asked "This was shot with the Red right?"

I turned around with my jaw on the floor. Then looked at the screen. Well... it was soft but it otherwise it looked excellent. There was no moire in that shot. Imagine his surprise when I told him "No, this is with the 7d". He was shocked.

Now granted, it doesn't always look that good. But there are times it shines if the DOP knows what he is doing.

Like this. An audience member may not be able to tell, okay, cool. Pick your priorities: my priority isn't if the audience can tell, it's peace of mind when shooting and knowing I'm not gonna run into any issues trying to capture an image to display for an audience member.

 

[dtjohns1]

My op was not saying that the hack is useless. I love the hack and can definitely see a difference between non-hacked and hacked footage. I'm just saying that I feel like after a certain point their ceases to be any changes going on. I would really like to see what everyone see's as the highest quality hack available right now so I can do a test for myself (great more testing now I'm being a hypocrite). When I did a personal test I used seaquake vs. brandin 66mb and didn't see anything different. I'm not saying I could have been wrong for having the wrong patch so that's why I'm asking for the top patch suggestions?

 

[mpgxsvcd]

Thank you in advance for your systematic work, mpgxsvcd, this is the kind of testing that provides genuinely useful benchmarks for making objective judgements. As other well-publicized shootouts have shown, contemporary video cameras have unexpected strengths and weaknesses. It's always advantageous to understand the limits of your tools in order to make informed choices in how you use them.

In addition to increasing encoder bitrate, PTool provides a wide variety of ways to fine tune the performance of the GH2. Compared to the GH1, whose issues primarily involved maximizing both bitrate and stability, the GH2 presents a far more complex optimization challenge. With the recent release of Panasonic's v1.1 firmware, supporting 1080p25 and 1080p30 AVCHD modes, we're starting a new phase of patch development which will require another round of careful testing before producing conclusive results. The clearly visible evidence we've seen of consistently measurable improvements in quality is an impressive achievement for this early stage of development.

Thank you for all of the work you have done. I don't think there ever would have been a GH2 hack if without the work you have done. Do you have any idea when the V1.1 hack will be ready? Is the new firmware significantly different? I am just curious about what the process is that has to happen before the new firmware can be hacked.

 

[Kholi]

My op was not saying that the hack is useless. I love the hack and can definitely see a difference between non-hacked and hacked footage. I'm just saying that I feel like after a certain point their ceases to be any changes going on. I would really like to see what everyone see's as the highest quality hack available right now so I can do a test for myself (great more testing now I'm being a hypocrite). When I did a personal test I used seaquake vs. brandin 66mb and didn't see anything different. I'm not saying I could have been wrong for having the wrong patch so that's why I'm asking for the top patch suggestions?

Quantum V2. That's what I'm on. Probably wont' go any higher or change unless it's to try GOP3 for cadence.

And, I have no issue with your OP. I think the returns have diminished greatly. My issue is mainly in the assumption that because someone didn't offer up double-blind tests or show everyone else what they're seeing (which they might not be able to, to begin with) then it's invalid or wrong. If someone doesn't see the difference, that's fine. That happens. But a ton of us are obviously seeing a better image, especially versus stock settings.

 

[Lpowell]

Do you have any idea when the V1.1 hack will be ready? Is the new firmware significantly different? I am just curious about what the process is that has to happen before the new firmware can be hacked.

In addition to minor improvements, Panasonic's GH2 v1.1 firmware adds two new HBR (high-bitrate) AVCHD video modes - 1080p25 for PAL and 1080p30 for NTSC. If you're familiar with the extensive list of PTool patches specific to 24p mode, you can imagine how much work it will require to track down, implement, and test corresponding patches for 25p & 30p modes. Vitaliy has been working full-time on this project since v1.1 was made available, but has not yet estimated a release date for the first V1.1 version of PTool.

 

[mpgxsvcd]

In addition to minor improvements, Panasonic's GH2 v1.1 firmware adds two new HBR (high-bitrate) AVCHD video modes - 1080p25 for PAL and 1080p30 for NTSC. If you're familiar with the extensive list of PTool patches specific to 24p mode, you can imagine how much work it will require to track down, implement, and test corresponding patches for 25p & 30p modes. Vitaliy has been working full-time on this project since v1.1 was made available, but has not yet estimated a release date for the first V1.1 version of PTool.

Do we expect the 24p patches and settings to operate any differently for 1080p 25/30? I wouldn’t imagine it would be any different. 24p to 25p is not a significant change in frame rate.

I guess what I really don’t know is what is involved in tracking down these settings? Are these settings all over the place and poorly named or is there real structure to how it is coded?

 

[Oedipax]

Out of curiosity, why are some people sticking with Quantum V2 rather than V5?

 

[Kholi]

Out of curiosity, why are some people sticking with Quantum V2 rather than V5?

My cards are limited. 30mb/s Sandisk. When I return after the end of the year I'm gonna try to land two 64gb 95mb/s so I can use the highest settings and span. I'll switch over to V5 then.

 

[Oedipax]

Oh okay, I'm just running the 30mb Sandisks as well. Haven't had any issues although I only shoot 1080p24 FHD so maybe there are issues at 60p I'm unaware of...

 

[OldCorpse]

Well, isn't it really a question of exact conditions specific to a given hack? I mean, if you take an unhacked GH2, and a hacked one, but the footage you shoot is full daylight uncomplicated image with few textures, you may not be able to see much difference. But perhaps a given hack shines in very specific conditions - like the blacks in low-light situations. Or a hack may do better with blowout rolloffs etc. Then there's the post - the original footage may look identical, but the hacked one can be CC'd much better and provides that flexibility. So it's important to see explore what the camera is capable of, so when you shoot, you play to the strengths of the camera in a given circumstance, and then a hack might be useful or not. That's a lot of tests.

 

[mpgxsvcd]

Well, isn't it really a question of exact conditions specific to a given hack? I mean, if you take an unhacked GH2, and a hacked one, but the footage you shoot is full daylight uncomplicated image with few textures, you may not be able to see much difference. But perhaps a given hack shines in very specific conditions - like the blacks in low-light situations. Or a hack may do better with blowout rolloffs etc. Then there's the post - the original footage may look identical, but the hacked one can be CC'd much better and provides that flexibility. So it's important to see explore what the camera is capable of, so when you shoot, you play to the strengths of the camera in a given circumstance, and then a hack might be useful or not. That's a lot of tests.

Yes it was a lot of tests. About 3 batteries worth to be precise. :laugh:

 

[DrDave]

If I get close enough to my 1920x1080 monitor then I can tell more from my computer screen than I could from the largest big screen out there. It is all about matching the display to the source resolution of the footage and getting as close to the display as focusing will allow. You don’t need a big screen to examine the footage. Your monitor will do just fine as long as it is 1920x1080 for 1080p footage and 1280x720 for 720p footage.

Have to agree with this. Looking at a calibrated 30 inch IPS screen from 20-24 inches away is the equivalent of screens that are much larger than your average movie theatre, plus the monitor is color corrected. I'm pretty sure I can see more detail on my monitor than on the best screen I ever saw in a theatre, which was the Skywalker Ranch screen. It's a very different feeling. I *prefer* the big screen for watching a movie, but I certainly can't do any pixel peeping. Which is good.

 

[DrDave]

Who here had an HVX200, HVX200A, and HPX170?

I had all three cameras. I used all three extensively. Of each, hands down, I swore up and down that the 200A looked better than the HPX170. Something happened in between the 200A and 170 that wasn't supposed to, it was noise and a color shift, I can't remember exactly what but they were supposed to be the exact same image. A ton of people didn't notice any difference at all, but there was just enough of a subtle change that it drove me mad.

I sold my HPX170 faster than I've ever sold any camera before. My HVX200A's image was great, but it only took that inkling of a difference to change my mind on the 170. To me, the Hack is the same way. There's enough of a change in the overall IQ, which could all be attributed to less compression and seeing more of the camera's true nature, to make me want to use it instead of tools that are obviously technically much much better. It may not be rocket science, but there's nothing that could've shown me different had I not picked the camera up before and after and looked at it myself.

The claims of increased DR, color accuracy, so on and so forth aren't what I'm looking at. It's strictly the fidelity of a captured AVCHD stream, which mainly comes down to the encoder's ability--when hacked, to retain detail as the camera shifts, or anything in the image moves around. It's also much MUCH more robust in lowlit situations, where lowlit's defined to me as having only five thousand watts of light, and I've got to go and light a large living room space with that.


I understand that some people do, and some people can't immediately see the difference in footage...

I'm not trying to be difficult, but I always come up against the same question again and again in my own tests, which I randomize. I'm not talking about audience members (although that is important, maybe even more important), I'm talking about the videographer.
If what you say is true, why can't anyone identify the cameras in the Bloom shootout? Why can people only identify things that are already labelled?

 

[Kholi]

I'm not trying to be difficult, but I always come up against the same question again and again in my own tests, which I randomize. I'm not talking about audience members (although that is important, maybe even more important), I'm talking about the videographer.
If what you say is true, why can't anyone identify the cameras in the Bloom shootout? Why can people only identify things that are already labelled?

Nah conversation is good. No prob. I think it's because we all come up with our own images in our experiences, right? Many MANY people were fooled by the green cast of the other images, because that's what most of us get out of camera. Bloom's findings didn't line up with what we were experiencing. If he actually got what we got, then perhaps it wouldn't have been such an issue for us to pick out the GH2 hack still from the rest. It's like when people post RED footage and it looks like trash. That's not what I get out of my RED footage, even in the worst case scenario. Why? Because I've been using it since the very first beta crop hit the streets and I know it inside and out.

Motion picture is also a huge tell. Stills are a different story. Have someone setup a shootout like the one Zacuto hosted, with really well lit scenarios, HUMAN BEINGS IN FRAME, motion, etc. You do a double blind like that, balance the cameras like they did, the post image, then let us go at a double blind. The results in that test were frighteningly obvious, and I don't believe that it would've taken much to discern which one was which.

To me, what is important is piece of mind. This is why I reach for a RED when I have to do a serious job. Being burned by Canons and the image was too much, I'd rather have a heavy camera and an image I know I can rely on instead of using something that's crippled just because it's cheaper and "the audience can't tell."

For the first time in a very very long time, I feel comfortable not reaching for an MX or an EPIC for every job. That kind of sentiment is not something I can put up and show people, it's only something that you know from countless hours spent on set, in environments that demand certain results and are pressure-intense, working with these tools. Again, if others don't see it, that's cool! I'm good with that. But obviously something is triggering an entire lot of us to jump on the Hack Bandwagon, and it isn't placebo effect.