Which screenshot is compressed? I dare you :)

[stream]

So I'm using ReMpeg on the very highest quality settings to transcode movies. I've found that no one I know has been able to tell the difference between the original and the compressed copy, but I figured this site would be a totally different ballpark. Here's some info first:

The compressed copy was scaled to 83% of the original bitrate. It took ReMpeg about 23 hours to transcode, running on a dual-933mhz processor machine. Pay no attention to the brightness difference between the two screenshots, as it really has nothing to do with the quality (ReMpeg apparently changes the brightness level a little when it transcodes -- but I won't tell you whether it's darker or brighter, of course ;)). Also, these screenshots were taken from the following source: JVC Stand-alone > Hauppage TV Tuner > WinTV2000 > Paint Shop Pro, with no compression on the .jpg files. The "stand-alone > TV Tuner" stage is why they look so grainy for DVD quality.

Anyway, here we go. Guess which one is compressed, top or bottom.

http://www.magnolia-net.com/~jnsb/1.jpg

I'm really curious to see if people can tell.

-jesse

[Faustus]

bottom I would say bottom, just a guess based off a few elements...

Screenshots can look good or bad depending on the screen, its how it looks on the screen that counts.

[jarvis1781]

yeah I would have to say the bottom too because there are a few more atrifacts running through her blouse area...but just a guess...honestly it doesnt matter to me which image I would be watching on a TV...both are good enough for me

[jarvis1781]

on second thought....you said the cap could have made the picture grainy...then I have to go with the top image as being the recompressed image....do tell me

[defense]

I just wanna say first off that I HOPE it isn't the top picture which was transcoded, because the "red" color in the girls face is terrible. Way off what it should look like. The bottom picture looks better to me in terms of "color," however, if you look at the entire background of the bottom picture, you can see that it is a little more blurred then the top. That is why my guess would be that the bottom picture is the one that was transcoded. I have not used ReMpeg, but I feel pretty confident that the bottom is the transcoded picture.


Also, you might want to conduct some more tests with movies that have high actions scenes. That is basically a "STILL" shot and I would think with ANY decent reencoder, you should be able to get a good output. Test some heavy action scenes and post the same test.

Is that scene from Vanilla Sky?

[stream]

Interesting, guys. Keep it coming, and thanks. If you want to know which one is which, just shoot me an IM.

Yeah, it's from Vanilla Sky.

-jesse

ps - Remember, the color changes you guys are seeing is really just a difference in brightness. I could have brightened or darkened one of them in PSP by just a smidge, and they would have looked identical in terms of "color." Defense, the red spots (I assume you mean right under her eyes) on the top picture are also visible in the bottom picture, they're just not as dark. Like I said, the transcoding process evidently changes the brightness of the picture. Also, thanks for the suggestion about using an action scene. I'll do that the next time.

[Adder_78]

You definetely should take picture from high action scene/scenes... There has been lots of these comparing issues like ReMPEG against CCE and allways the CCE has win... I dont see any point of using ReMPEG but thats just my opinion

[dhluke]

With a picture of that size it's hard to tell... Do a bigger resolution for the picture and I'm sure it'll be noticeable which one it's the copy or not...

dhluke

[BJ_M]

unless there is movment you are just basicly doing a jpeg compression on a still shot

[Faustus]

I guess we dont get an answer?

[vhelp]

.
.
sorry, but there are too many variables to make a choice

Movement will tell/give more trueth to a "tell the difference" scenario.

Until then, your "still" pic is flawed.. maybe because the MPEG at a steady
scene w/out motion (or much) didn't require ReMpeg to do anything, if
any at all.

And, forget the PIC unless you must, instead, U/L an actual MPEG for such
a test like this. PICs are usually useless unless demonstrating artifacts in
things like filtering and such

-vhelp

[SaSi]

The bottom picture also suffers from horizontal artifacts that seem to be caused by the player or capture method (would not be caused by bad compression).

All in all, these two pictures are not comparable (at least so that any subtle difference caused by an additional encoding step can be identified).

A point there is that even if a video is re-encoded, if it's done decently, several other parameters will affect more than the additional loss in quality.

[ZippyP.]

To my eyes the bottom picture looks sharper but I like the top one better. I think the top one is the compressed one.

[dhluke]

Download the picture and then use whatever program you use to open pictures and zoom in them -- you'll notice the blocks on the wall and also you'll see artifacts on her blouse... hmm and no we aren't perverted!


dhluke

[jacknjt]

the top one is the compressed for sure. because it shows less artifacts that mean the encoder wiped away some facts from the pic.

[red lion]

I think the bottom one looks more blocky in areas but i would not be surprised if they both were compressed just to trick us and probebly from the 2 different progs that everybody is fighting about to prove there is not much difference
just my guess

[stream]

The top is compressed, the bottom one the original.

-jesse

[LiquidXTC]

woooo

[bbb]

Bottom one is original straight from capture card (I can see the yellow horizontal line interference that comes from TV Tuner capture cards; also TV Tuner cards tend to capture very brightly).

Top one is the compressed one which has been noise reduced (looks like SmartSmootherIQ), color corrected (although the skin tone looks too pink/purplish) and darkened. Also, the facial features are a little bit blurry which indicates touching up/filtering.

[stream]

bbb:

Actually, both shots are simple screen captures of my TV Tuner's wintv2000 program, then cropped. I used PSP 5 to get the captures. Also, I didn't edit either of the shots in any way. They're both totally untouched. I captured, cropped out my desktop, etc. and then saved to .jpg with no compression.

-jesse

[flaninacupboard]

the discussion is pointless, they're not full resolution, you've captured them instead of just taking a frame out of the mpeg (which is just more work for yourself, and not very accurate) and you say no compression has been used in the jpeg, then it's clear you don't even understand what a jpeg is.....

[stream]

Well, by "no compression" I mean it was set at the "highest quality" possible .jpg

-jesse