Canon 5D Mark III -IPB contains more detail and has less artifacts than ALL-I

hoho:smile: my test are like those in the link, anyway i don't watch my work in 100% crop, who does? what i trust is what i see clean right before my eyes, graded or not
 
nikon_obsessed said:
hoho:smile: my test are like those in the link, anyway i don't watch my work in 100% crop, who does? what i trust is what i see clean right before my eyes, graded or not
Click to expand...

Do you check your work at 100% on an HDTV, etc.? When we have issues with image quality, we study at 100% or more (cropped on a computer monitor, fullscreen, or on an external monitor (HDTV, etc.)). If you compare ALl-I to IPB, at first ALL-I can look sharper and more detailed. That's due to aliasing artifacts on macroblock edges and additional noise. After sharpening, grading, and adding custom organic noise, IPB is looking better in my tests: more filmic as it doesn't have aliasing artifacts from macroblocking, and I can carefully tune & control the noise based on content (unlike the baked in ALL-I noise).

If you could post a still frame from ALL-I compared to IPB which shows a quality issue you are experiencing, we can analyze it to try to find out what's going on.

I purchased a Lexar 600X Pro CF card to see if I could coax ALL-I past IPB in terms of quality. While at least one clip hit 101Mbps, it still had macroblocking and aliasing issues (along with baked in noise). I will continue to try ALL-I in different situations as I shoot my movie, however I'll always cover the same shot with IPB for safety.
 
I have 7 TV's. Here are the sizes. 200", 72", 60", 2x 47", 2x 37". My mk3 footage looks great on all of them. As well as my af100, xf105, gh2, even my 5dmk2 if not shooting bricks.
 
well, got no issues with all-i. here you are given a more superior and robust codec and yet i can't imagine shooting with an inferior and highly compressed one, don't find ipb more filmic or less aliased. like i said, i don't woo my client with a 100% crop, they'll probably think i'm an idiot, what they saw on the monitor put a big smile on their faces. if i get a mk3 to shoot ipb, i might as well save my dough and stick to mk2. even the fellow who started all these fizz nonsense is advocating ALL-I.
 
dustylense said:
I have 7 TV's. Here are the sizes. 200", 72", 60", 2x 47", 2x 37". My mk3 footage looks great on all of them. As well as my af100, xf105, gh2, even my 5dmk2 if not shooting bricks.
Click to expand...

That is a remarkable viewing collection! So nice to see footage large like that.
 
Cosimo Bullo said:
That is a remarkable viewing collection! So nice to see footage large like that.
Click to expand...
My point is not my collection. It's saying that if it looks good on my 200 and my 72, then it just looks good. And that is what matters. On a computer screen, I see all that JCS is talking about when I zoom 200%. Plug your camera into your favorite TV you have with HDMI. Tell me what you see...

@nikon_obsessed... When you watch a TV, you're in fact watching material at 100%, be it on a 200 or a 72" or even a 12" set.
 
nikon_obsessed said:
well, got no issues with all-i. here you are given a more superior and robust codec and yet i can't imagine shooting with an inferior and highly compressed one, don't find ipb more filmic or less aliased. like i said, i don't woo my client with a 100% crop, they'll probably think i'm an idiot, what they saw on the monitor put a big smile on their faces. if i get a mk3 to shoot ipb, i might as well save my dough and stick to mk2. even the fellow who started all these fizz nonsense is advocating ALL-I.
Click to expand...

ALL-I makes sense for editing speed, but it's not efficient. Canon's C300 uses a GOP codec based on MPEG2 (ancient, but highly optimized). At 50Mbps 422 (twice the color information of 420 for the 5D3), it produces excellent quality (much better than ALL-I on the 5D3). H.264 IPB is the most advanced and efficient codec currently available (ALL-I is effectively MJPEG with variable macroblocks). Not only does Canon's ~30Mbps IPB outperform ALL-I, so does Nikon's 24Mbps codec on the D800 (tested by Dan Chung- the footage is clear). I have posted crops showing the issues with ALL-I on PPro CS5.5.2 (less of an issue on OSX with Compressor, etc.). I have not found a case where IPB performs worse than ALL-I (please post if you find such a case). If you can't see any loss of detail or additional artifacts with IPB and have fast computers to edit with, IPB will save ~3x the disk space and allow ~3x recording time.

In my day job I write image processing software; work with codecs at low level, etc. Folks found issues with ALL-I and asked me to investigate and share the results. Until someone posts clear examples showing IPB performing worse than ALL-I, I recommend IPB when max quality is desired. I understand that it might be hard to believe a ~30Mbps codec can match or exceed a ~91Mbps codec, however H.264 IPB is the most technologically advanced codec on the market. The disadvantage of IPB is it requires a powerful computer/GPU to decode in real-time. Moving forward we'll see the high-end HD and 4K cameras using 422, 444, 10-bit, etc., H.264 IPB, and they will look awesome while being fairly efficient. Look at the quality of 50Mbps MPEG2 from the C300- it's fantastic. H.264 is much more efficient than MPEG2: the new H.264 formats coming out will look even better.
 
Last edited:
nikon_obsessed said:
different cliente i guess, 100%,200%,1000%, 200 or 72" crop or what doesn't interest my client. good for me.
Click to expand...

Folks reported problems with ALL-I which I have verified and posted video and stills. Some stated IPB looks worse than ALL-I: I haven't found that to be the case after analysis. When asked for screen shots showing issues with IPB, the discussion changes to ALL-I looks good enough. The scientific method is useful in problem solving, however there is also a psychological component: I feel it as well- I want ALL-I to work better as it edits smoother on laptops, etc., and any way to improve quality of 5D3 footage is welcome. It should be possible to get ALL-I closer to IPB via a firmware update (less noise) as well as improved decoders (proper DCT decoding, proper colorspace handling, and improved deblocking).
 
good thing the folks i work with aren't interested in engineers, just the typical show me nice stuff type, they loved the clean ALL-I, even better when grading it. test charts bore most clients.
 
nikon_obsessed said:
u mean your cilents are blown away by amazing test charts?
Click to expand...
Dude, first welcome to the forums. Second, out of the 10 posts you have contributed, all 10 are combative. Just saying.

Now, I can't think of a camera that people have not pointed at test charts when first released. Why not thank the people that are doing it for you? They might discover a limitation of a camera through their tests. They might show you some good starting settings for the camera you just bought but are too lazy to point at test charts.

I have been working with the mk3 for 3 weeks now. It's paid itself off easily. But, i still read what JCS and others have to post. Sure, my clients are very happy with the footage I've shot with it. But, I still tinker with shooting different patterns along with different lenses to "know" what to expect. It's kinda like the old days of testing film, then labs, then lights and temperature. I have at least 60 light sources in my stills kit. They are all color balanced to each other. They range from $25 battery strobes to $1200 strobe heads. All of them have filtration to match each other within 50 kelvin. If one light starts showing a color shift, I see it immediately and take it out of production until I correct it. I had to test for that. It's part of photography and no different than shooting test charts. A good shoot has to understand his tools. My clients didn't request it, but I did it for my own benefit. Photographers test, it's just as simple as that. Nice people share what they have found, although I keep some secrets to myself.

So, maybe simmer down and "join" the community rather than combating users here for sharing info they have found, wether it's opinion or fact. You might just learn something.
 
what dustylense said

and "All-I looks good" is not a reason to choose it over IPB, if IPB looks even better
(lucky you, that are choosing between good and better...)
 
Part of the issue with ALL-I at the time of the original post may have been due decoder quality with Premiere Pro CS5.x. If you can see higher quality with ALL-I vs. IPB in your own footage it makes sense to use it. If not, no need to waste ~3x the disk space. Premiere Pro CS6 works very well with IPB and ALL-I in both Windows and OSX, desktop and Macbook Pro (late models with NVIDIA GPU accel).
 
You must log in or register to reply here.
Back
Top