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

[Kholi]

You guys have this compression thing wrong. You do not lose quality just because you edit footage. That's what high profile formats are for. A proper conversion to ProRes HQor 4444 is pretty much lossless for the first and second generation. DnxHD the same.

You only run into issues exporting for exhibition and it's simply a lack of knowledge on the part of the person doing so that is the weak link, not the idea that you can't retain detail. You will always, always end up with a better image when compressed from a good image versus bad image base then compressing it. There's no magic. It's just logic.

You do not lose advantages of the hack, that is wrong. Learn how to compress for you exhibition format. And again if yore having trouble understanding this just look at Tonaci Trans scarlet footage on vimeo or Apple Movie trailers.

 

[Sage]

Excellent

 

[mpgxsvcd]

You guys have this compression thing wrong. You do not lose quality just because you edit footage. That's what high profile formats are for. A proper conversion to ProRes HQor 4444 is pretty much lossless for the first and second generation. DnxHD the same.

You only run into issues exporting for exhibition and it's simply a lack of knowledge on the part of the person doing so that is the weak link, not the idea that you can't retain detail. You will always, always end up with a better image when compressed from a good image versus bad image base then compressing it. There's no magic. It's just logic.

You do not lose advantages of the hack, that is wrong. Learn how to compress for you exhibition format. And again if yore having trouble understanding this just look at Tonaci Trans scarlet footage on vimeo or Apple Movie trailers.

So how are you going to upload a 10 minute lossless file with a 2 GB cap? Somewhere along the line you have to recompress it to streaming bit rates. You will lose the benefit of a high bit rate original unless you are just trying to do something to introduce compression like waving the camera around.

Check the files on Vimeo. I couldn’t tell a difference once they are at streaming bit rates.

 

[sparky33]

This is exactly why I sold my AF100. I absolutely loved it, but got sick of hearing from my wife "Oh, you taking THAT big thing out with us again?"

Easier said than done, Sir!

I only do video as a hobby - Steam Fairs, family occasions, that sort of thing.

When I had my little HG10 all was serene - I'd produce the camera as we walked around, shoot some footage, pop it away, and all was well.

However, as I added to my GH1/2 colection - rig, handles, a external recorder, external monitor, etc, the family became ever more reluctant to be seen in my company

The final straw was the Matte Box - which may have broken the proverbial camel's back, but most definitely not it's spirit!

Simmering discontent has become open rebellion - "You're *not* coming around with us with *that* thing...."

Oh, I've tried to explain what the matte box is for - tried to tell them about the necessity of using ND filters to keep the shutter speed at 50/1000 - but they just don't care!

Bottom line? - they think I'm walking around like Stephen Speilberg, and they think it's highly peculiar.

Perhaps it is .....

 

[Kholi]

So how are you going to upload a 10 minute lossless file with a 2 GB cap? Somewhere along the line you have to recompress it to streaming bit rates. You will lose the benefit of a high bit rate original unless you are just trying to do something to introduce compression like waving the camera around

You do realize that the average Apple Trailer is about 2:30 Minutes and the 1080 Files are typically around or fall under 200mb's right?

Do you see them "losing" the advantages of shooting superior formats? No, because again, garbage in, garbage out. If you learn how to compress your footage for your viewing format then you won't have the issue. And, you can't say it's because of YouTube, either. The Hobbit Trailer looks damned good @ 1080 even on Youtube.

Check the files on Vimeo. I couldn’t tell a difference once they are at streaming bit rates.

If they werent' compressed well to begin with, then looking at them won't really show a difference.

 

[DBP]

Here's an example of what Kholi is saying.

Check out this test.

Watch it in 1080p and fullscreen it. The compression artifacts look considerably worse at the stock and lower bitrates. Youtube tends to recompress to around 7-10mbps at the most, regardless of the source quality. I've done tests. My last 12 minute short film, I uploaded a 9gig 100mb h.264 file, and the youtube recompressed file was still about 10mbps.

Point being, higher bitrate source files come out of youtube/vimeo compression much better. Maybe not at 360p in a window. But at 720 or 1080p fullscreen, the difference is definitely there, in my experience. Whether this is a big deal to you, well that's up to you.

 

[Danielvilliers]

Look at my example in a post on personal-view on the dynamic range of the hacked gh2 vs 7d http://www.personal-view.com./talks/discussion/1795/dynamic-range-test-between-gh2-and-7d/p1 . Now for those that cannot see the files look here the original file http://www.personal-view.com./talks/uploads/FileUpload/66/84806be35961068f655d09bcb5a2ba.png one where I flatten the look http://www.personal-view.com./talks/uploads/FileUpload/fe/12188d70e90e5b45a9dbec331975c8.png and a last one where I boosted the shadows aggressively while maintaining the contrast level http://www.personal-view.com./talks/uploads/FileUpload/dc/22f554fb615dff9dd63e9dad1ef514.png (The numbers is to calculate the number of stop of the scene). Unfortunately the aim wasn't to test a hack vs stock gh2 footage. With my experience I can tell you that the stock ones would have shown a lot of macro-blocking at these level. Look everywhere and you will see how everybody was reporting so on the stock gh2 footage. Yours and mine are static scene (tripod and non moving character etc), in motion I sure it would have really shown.

The reason is simple, these codec are meant to compress everything that is... invisible in theory. But this scene is beyond the DR of the curve/profile that was use from the RAW. A flatter profile (unfortunately you cannot set the idynamic range, which seems to compress/fit a bit more of the DR of the sensor before encoding, some thread discuss it on personal-view, with some test, but it can bring a little more noise but still better than boosting the shadows in post after they have been encoded). So the hack does not mechanically increase the DR of the gh2vk. It just renders a more accurate image from the highlight to the shadows because it has more bitrate to work with. It is the same case as with high Iso, the improvement is that the noise is better preserve, which might look as strange thing to say. But the noise is better preserved, so you get better detail, which for many look like fine film grain and to others can be cleaned much more easily.

 

[Ian-T]

I still contend that there is no discernable compression difference in the original files.

But why would there be in this example? We’ve seen plenty of videos by others that show the advantages especially in high contrast dark areas (grass, trees, fine detail in the shadows etc). Then again....those are all moving shots I am referring to as opposed to these stills.

I think you are really misguided on what you are saying. The hack does not affect dynamic range at all. The test files are there and not a single person has shown any differently. There are differences but dynamic range is not one of them.

I’m sorry but it does affect “useable” dynamic range. The bottom line is it’s about how well the hack fills the “available” dynamic range (whatever that may be) with useful information. When viewing static images it will not be apparent. But when one is color correcting/grading it slaps you in the face. That in itself is a direct affect on the “difference between the smallest and largest possible video brightness”...or in other words...”dynamic range.” Un-hacked images on this camera just smears away valuable information in the dark parts of the image.

The whole point of the video was to give everyone a chance to judge for themselves if they saw a dynamic range difference. I would love for someone to show a difference in that. Until then I will leave the video as is.

Because of the efficiency of AVCHD I think that would be almost impossible with a still from video. This is just not a good way to test what you are trying to prove (or disprove) IMO. To me....it’s almost like a trick question....which is somewhat misleading.

 

[DrDave]

So how are you going to upload a 10 minute lossless file with a 2 GB cap? Somewhere along the line you have to recompress it to streaming bit rates. You will lose the benefit of a high bit rate original unless you are just trying to do something to introduce compression like waving the camera around.

Check the files on Vimeo. I couldn’t tell a difference once they are at streaming bit rates.

This is the end user reality. It is like in still photography when pixel peeping, you print it out at different sizes to see what it actually looks like. And some of the quality comes through, and some of it is left behind. Even if we have a jumbotron in our home, most people will view it "downstream" even in a theatre. That doesn't mean we shouldn't make a reasonable attempt to future proof the master at a very high quality, but no one may see it. If I have a client over to preview a work on the big screen, they are always shocked at what it looks like. I explain that they can download a pretty high res copy from Vimeo, and play it on a media player, and maybe one person has actually done that, and three more people have looked at a blu-ray disc I burned for them. So what is that, five people out of seven million? Of course quality is important. But many won't see the top resolution--most theatres even have crappy screens and blown out speakers.

 

[mpgxsvcd]

Here's an example of what Kholi is saying.

Check out this test.

Watch it in 1080p and fullscreen it. The compression artifacts look considerably worse at the stock and lower bitrates. Youtube tends to recompress to around 7-10mbps at the most, regardless of the source quality. I've done tests. My last 12 minute short film, I uploaded a 9gig 100mb h.264 file, and the youtube recompressed file was still about 10mbps.

Point being, higher bitrate source files come out of youtube/vimeo compression much better. Maybe not at 360p in a window. But at 720 or 1080p fullscreen, the difference is definitely there, in my experience. Whether this is a big deal to you, well that's up to you.

Thanks you just proved my point. That is actually a perfect example of how the original quality doesn’t matter. Take a look at the light from the flame in the arch just below the Eifel Tower and in its shadow on the wall. In the scenes where the flame is flickering there is obvious compression in the flickering light because youtubes compression can’t handle it. In the scenes where there is no flickering the compression looks identical even at 1080p.

The compression shows up for both the high bit rate hacked footage and the low bit rate unhacked footage. It only depends on whether the flame is flickering or not.

Actually you can right click and select video properties and it will show that this particular video streams at 2.32 megabits/sec. I think it will go higher for the “original” quality options. However, it never goes much higher than about 4 megabits per second. That is very low and can easily show compression artifacts from any source.

 

[mpgxsvcd]

Because of the efficiency of AVCHD I think that would be almost impossible with a still from video. This is just not a good way to test what you are trying to prove (or disprove) IMO. To me....it’s almost like a trick question....which is somewhat misleading.

Who said anything about stills. I posted the original unedited files that everyone can download from Vimeo. I am talking about comparing those videos not just stills.

 

[Danielvilliers]

Still... I think it is obvious that it meant on tripod and no movement even in the scene, I won't go and search for the test ralph-b did on personal-view comparing hack (in this period it was the 132mbit 3 gop driftwood hack that was the highest) with the hdmi uncompressed output, and even the 44 mbit hack was significantly better than stock. I think everybody has there own belief, so be it and if they want they can search for it.

I was bred in the advertising market for 20 years where you always go for the highest quality even if most of it ended on news paper on 85 lpi, where you could see clearly see the quad dithering on the colour separation film. But the principal of junk in, junk out stands, because sometime it could end on magazine glossy paper. So you always aim for that quality because you never know where it is going to end and always be prepared for that.

I apply the same principal for my filming. I have no pretension that my films will go on the big screen but I will aim for that. So get the best quality in case it happens and for my use and self promotion it is always good to have the best quality to showcase my work. Because the people that are going to judge my work are the same as me. They won't judge me as the same level than joe public, because they are in that domain and I will have to pass their scrutiny... and the competitors.

I always look at a film I always look at the story, the acting but because of my specialty as an image maker, I will also concentrate on the framing, lighting, grading etc... Because that is why I am paid, not what joe public can do but beyond that level, if else why should they hirer me in the first place.

It is true that once in a while a film done on hdv becomes a big hit, but they are more an exception than confirm the rule. If else there would be no place for the Epic and Alexa of this world.

 

[mentatDUKE]

It's funny seeing people still debating this. What part of garbage in garbage out do people not understand. We simply want a high quality master.

 

[Kholi]

I was bred in the advertising market for 20 years where you always go for the highest quality even if most of it ended on news paper on 85 lpi, where you could see clearly see the quad dithering on the colour separation film. But the principal of junk in, junk out stands, because sometime it could end on magazine glossy paper. So you always aim for that quality because you never know where it is going to end and always be prepared for that.

I apply the same principal for my filming. I have no pretension that my films will go on the big screen but I will aim for that. So get the best quality in case it happens and for my use and self promotion it is always good to have the best quality to showcase my work. Because the people that are going to judge my work are the same as me. They won't judge me as the same level than joe public, because they are in that domain and I will have to pass their scrutiny... and the competitors.

I always look at a film I always look at the story, the acting but because of my specialty as an image maker, I will also concentrate on the framing, lighting, grading etc... Because that is why I am paid, not what joe public can do but beyond that level, if else why should they hirer me in the first place.

It is true that once in a while a film done on hdv becomes a big hit, but they are more an exception than confirm the rule. If else there would be no place for the Epic and Alexa of this world.

It's funny seeing people still debating this. What part of garbage in garbage out do people not understand. We simply want a high quality master.

I can't understand why this concept is escaping people so very easily. I really cannot. With Fibre Optic going into homes strong (4G streams) maybe people will start getting it when we can get a superbly compressed 6mb/s MPEG-4 stream into homes, right to televisions? ... Probably not then, either...

 

[JMZ]

I can't understand why this concept is escaping people so very easily. I really cannot. With Fibre Optic going into homes strong (4G streams) maybe people will start getting it when we can get a superbly compressed 6mb/s MPEG-4 stream into homes, right to televisions? ... Probably not then, either...

If the high bit rate hacks are in fact bogus why does anyone care? Meaning, if people get out and shoot more things and do it well does it really matter if it was shot with a full sensor, m4/3 sensor, low bit-rate, high bit-rate or Creole voodoo? As someone who shoots video and as an editor who's slowly working into 3D animation I'm fairly used to pixel peeping and encoding over and over to cram just a few more bytes into a deliverable. But I'll be the first to admit that sometimes it's really just poking around for the lack of anything better to do. Yet it's fairly universal in any kind of image-making (print, film, video, web) that you want a high quality original (as high as you can reasonably get) before you do any manipulation. Exactly because you know it will lose detail at the far end of the distribution chain. With all of the available hacks maybe we can think of it like digital film stock. I think someone already mentioned that in another thread.

Anyway, if the hacks are no better than (insert alternative here) and we are all suckers what's the harm? I maintain my wife's SanDisk mp3 player is better than my co-worker's iPod because it has many more features, a replaceable battery, is less likely to be stolen at the gym and was one fifth of the cost. The facts are there on paper and in real world use. But both people are happy with their mp3 players so in the end it doesn't matter. I'm all for people who have the time to test and retest what the cameras and hacks can do. I also appreciate those who want to test out hacked versus stock whatever in side by side comparisons. But for what I do...I want the absolute highest quality I can get without bricking my camera or frying my card.

I just don't get why it's so important to prove these things (med vs high rate hacks, hack vs stock, dynamic range improvements vs not, resolved detail) aren't being improved. I'm not trying to continue the argument or incite anyone to anything. I'm genuinely curious. What's the benefit? We can all agree the GH2 is great. Most of us (me anyway...) agree the GH2-X is even better. And some of us even agree that a high quality original is the best option for maintaining some control through post and beyond. If that's delusional then I guess we should get a group medical plan because more and more people are turning to the GH2 and becoming active in this forum and getting into shooting.

So... why don't we just encourage each other to shoot great stuff and help the community grow?

 

[mentatDUKE]

Why not do both? The arguments are necessary. Call it scientific peer review if you will. This happened in the early days of the 5D also. 5D Discoveries: Moire, Line Skipping, Aliasing, etc. From the early findings sprang Zacuto's disciplined way of doing things and from there sprang the SCCE. The debates happen, then tests are performed and the "truth" is revealed. The indie filmmaking tech community is essentially a think tank.

We don't have the ASC or the ISO to vet our digital filmstocks for us. The vetting is now basically community driven and is allowing $500 cameras to deliver million dollar images.

So this is all very necessary.
At this point I'm very satisfied with camera quality. Only a few things need to be fixed now before we have the perfect indie camera aka the $1 paintbrush aka true artist tool:
No more skew/jello cam.
RAW or 10/12-bit or 4:2:2 (but in a dream world, RAW)

My how tech changes. I started during the 35mm adapter era and here we are. All for a fraction of what I paid for my dvx/Redrcok rig. Amazing. At this point quality isnt a barrier and price isnt a barrier. Get busy livin or get busy dyin.

Looking forward to the new year...

 

[Ian-T]

Who said anything about stills. I posted the original unedited files that everyone can download from Vimeo. I am talking about comparing those videos not just stills.

My bad...I should clarify..."video images that have no movement in them...thus not stressing the codec one iota."

Even the one DBP posted a few posts back was a good one. That filckering candle screwed with objects in the shadows and gave away the good bad and ugly of compressed video vs. "less compressed.".

 

[mpgxsvcd]

My bad...I should clarify..."video images that have no movement in them...thus not stressing the codec one iota."

Even the one DBP posted a few posts back was a good one. That filckering candle screwed with objects in the shadows and gave away the good bad and ugly of compressed video vs. "less compressed.".

I agree any motion at all or any major noise can introduce compression artifacts. However, I intentionally eliminated that motion in my tests because it is too hard to control. If anyone wants to do a controlled motion test I am all ears. I would love to see how motion really affects the codec. I just time have the time or equipment to accurately test it.

 

[JMZ]

Why not do both? The arguments are necessary. Call it scientific peer review if you will. This happened in the early days of the 5D also. 5D Discoveries: Moire, Line Skipping, Aliasing, etc. From the early findings sprang Zacuto's disciplined way of doing things and from there sprang the SCCE. The debates happen, then tests are performed and the "truth" is revealed. The indie filmmaking tech community is essentially a think tank.

We don't have the ASC or the ISO to vet our digital filmstocks for us. The vetting is now basically community driven and is allowing $500 cameras to deliver million dollar images.

So this is all very necessary.
At this point I'm very satisfied with camera quality. Only a few things need to be fixed now before we have the perfect indie camera aka the $1 paintbrush aka true artist tool:
No more skew/jello cam.
RAW or 10/12-bit or 4:2:2 (but in a dream world, RAW)

My how tech changes. I started during the 35mm adapter era and here we are. All for a fraction of what I paid for my dvx/Redrcok rig. Amazing. At this point quality isnt a barrier and price isnt a barrier. Get busy livin or get busy dyin.

Looking forward to the new year...

mentatDUKE,

You make some very good points. I think testing and experimenting are great and agree it's a necessary component of the process. It will allow us to develop a set of tools we can all take advantage of depending on our needs. But lately (and this is just my opinion...) there seems to be a bit of an undercurrent aimed at disproving the things the hacks have brought to the table. Or perhaps simply saying they're not as great as we all might think. I don't really care because I'll test what I need, compare it to what other cams I have available and use the one that works best for me.

Maybe I'm still feeling after effects from other camera forums I used to frequent more often.

I dig your $1 paintbrush metaphor too.

 

[maarek]

You make some very good points. I think testing and experimenting are great and agree it's a necessary component of the process. It will allow us to develop a set of tools we can all take advantage of depending on our needs. But lately (and this is just my opinion...) there seems to be a bit of an undercurrent aimed at disproving the things the hacks have brought to the table. Or perhaps simply saying they're not as great as we all might think.

Speaking as someone who doesn't own a GH2 yet (gasp!) but who does do a lot of pixel peeping because my work duties are a lot of encoding/compositing/vfx work, I would say that the hacks enable something that I've always wanted. Intra-frame encoding. You basically get rid of all the compression artifacts that have been plaguing cameras since forever. Now 90mbit/s versus 170 is pretty much debatable. I'd personally go with the lower one just because of cards. But at those bitrates you can achieve good quality and diminishing returns after 100mbit/s.