Picture % Quality test results of Instant Copy

[mrbass]

Did it take you all day, all morning, remotely doing it from work, etc.? Well...glad you asked..as a matter of fact it did.

I did fire explosion and water explosion test at 90%, 80%, 70%, 60%. Anyway I might just do 75%..because that might be a sweet spot. At 90% it's pretty hard to tell the difference. Obviously this is action frames. I'll try doing same thing with Spirited Away soon and see if that looks good at 60% or so.

You can find the links to the tests at the very bottom of the guide.
http://www.mrbass.org/instantcopy/

[BlakIce]

couldnt see much difference, could be im sleepy, thanks anyway

[mrbass]

Ok well I guess it's hard on a webpage...I should've just zipped them all up. Anyway download the 90% one and 60% one. Then flip back (backspace) and forth (spacebar) with irfanview and then say you can't see a difference. I can tell a difference from the original compared to 90% but boy is it miniscule. My whole point of this exercise was to determine what is the minimum acceptable pic quality percent for action movies. For me I'm guessing it needs to be at least 70%. 75% might just be the sweet spot like I said before.

[MackemX]

the difference is there, and I would agree 70% is about the minimum for acceptable quality loss, as 60% in both pics looks the same as if someone (not you mrbass:)) had added a slight blur to the original saving as a jpg creating the blocky effect using Painsthop Pro

but I now think people's opinons of sample pictures are affected by how good the picture is displayed on their screen as I have posted here (http://forum.doom9.org/showthread.php?s=&threadid=46630#post266205)

[mpucoder]

Just want to throw something in for consideration. The same question came up with ReMPEG2, and it was my opinion that bitrate, and not percentage, is more important. A low bitrate original can't be reduced as much as one of those super high bitrate disks and look as good. And even high bitrates don't guarantee good source encodes.

[mrbass]

@MisterX
hmmm well I don't necessarily agree. I have a 19" inch flat CRT and macroblocks at 60% are quite noticiable and if I flip back and forth I can see differences between all percentages. I also have a 17" LCD 25ms 350:1 granted not DVI (but that's mostly to prevent ghosting caused by EMF emmissions which I don't seem to have and I've seen them on LCDs at work a few times).
Looks like I blurred it with Photoshop?

Ok here's what I did. I obviously had to do a 2 hour encode on the movie for each percentage. Appended the .pdi with .vob to preview it in PowerDVD. In PowerDVD set Color Control to Original, Keep Aspect Ratio..ScreenShot current video window size and capture to .bmp. I did a few tests to see which percent .jpg would start losing detail. I opened .bmp in irfanview and save the .bmp to .jpg at 75%, 80%, 85%, 90%....I concluded at 90% it was extremely close to the .bmp. At 85% jpg it was just a tad unlike the original (jpg to bmp).

Anyway my whole point is that I didn't blur anything or block anything...What you see is what Instant Copy Encoder/Transcoder did at those Picture Quality Percents...I encoded the whole movie each time in its entirety for each percent and screenshoted that and saved each screenshot at 90% .jpg.

@mpucoder
sounds reasonable....the original was a very high bitrate usually 9Mbps IIRC just so you know.

I guess when it comes down to it each and everyone has to do their own tests. For me if there is a wonderful bonus feature which I'd like to include but it drags down the main movie quality below 75% or 70% than I'll have to exclude it.

[MackemX]

Originally posted by mrbass
@MisterX
Looks like I blurred it with Photoshop? :D you misunderstood me, I didn't mean you in general, I mean't anybody, I will update my post again to give a better explaination

I agree you can see the difference on a CRT also, but my point is that the TFT will show a greater margin of quality between the original & 60%, compared to the difference between the 2 on a 17" or 19" CRT

p.s. surely you must agree that a 17" TFT displaying 1280x1024 shows a higher quality definition picture (be it minimal to most users, (i.e. my mother) than the CRT with same resolution setting if both monitors are of similar performance within their fields?, if you disagree then that must mean the only reason people buy TFT's is because they have a smaller footprint! (mean't as a friendly quip nothing else :) )

[mrbass]

ok ...I was just clarifying so one wouldn't think I was "doctoring" the photos or what have you. About the LCD....I'm not gonna get into that too much. My only point about the LCD is that if you have one or you don't you can still see a difference (and not a huge difference if both are at 1024x768 for example). If your saying LCD screenshot at 80% looks like crap then it's hard to gauge how each perceives crap cuz on my LCD and CRT it looks about the same. Anyway this LCD arguement could spark a few hundred posts so I'd rather not get into it here only to say that yes you win..hehe.

I think doom9's shootouts are excellent and help people judge for themselves so to speak.
MPEG-4 Codec shoot-out 2002 - a comprehensive test of all the latest MPEG-4 codecs
MPEG-4 Codec shoot-out 2002 2nd installment - as the above but featuring SBC, DivX5.01 XviD 20020327 and VP4
MPEG-4 Codec shoot-out 2002 3rd installment
http://www.doom9.org/doc-overview.htm

Also mb1's screenshots on his excellent guide helped me decide to try out instant copy for myself.