QUAKELIVE movie making guide by aNtii

[fraggedICE]

well, i just took a jpg screenshot and a tga screenshot. i opened them both in photoshop, zoomed in closely, looked from far away etc. and they were both the exact same, pixel for pixel. and i thought that i MIGHT have missed something, so i saved them both as .bmp (you said it was lossless and uncompressed, so what the hell...).

The .tga > .bmp: 4.94 MB (5,184,072 bytes)
The .jpg > .bmp: 4.94 MB (5,184,072 bytes)

however, the outputted .jpg is 1.79 MB (1,881,062 bytes), while the .tga is 3.70 MB (3,888,018 bytes). 1440x900, every single tga screenshot, no matter what, will be the exact same as the jpg, but it will be a much larger file. I didn't test /video, which I might do next, but this proves that .jpg = .tga, though i dont think it was part of the original argument. I guess ill test jpg/tga/avi next.

This is the exact same argument as before; you can open the .tga and .jpeg from my files and save them as .bmp, they are exactly the same file size, byte for byte.

Just like uncompressed .AVI files, BMP is lossless while JPEG and TGA are not. When you save them as .BMP it reads them pixel by pixel [QUOTE=KittenIgnition;104201]jpgs dont do that, and vary in size depending on how many colors etc. are in the image.[QUOTE]

When you zoom in and out you will probably not see any differences though, as QL uses a very high quality type of JPEG. There is some difference though, my method proves it; all JPEGs use the blocky JPEG compression but TGAs use another type of compression which is almost lossless which is why their file size is larger.

JPEG ≠ TGA

[KittenIgnition]

This is the exact same argument as before; you can open the .tga and .jpeg from my files and save them as .bmp, they are exactly the same file size, byte for byte.

Just like uncompressed .AVI files, BMP is lossless while JPEG and TGA are not. When you save them as .BMP it reads them pixel by pixel

When you zoom in and out you will probably not see any differences though, as QL uses a very high quality type of JPEG. There is some difference though, my method proves it; all JPEGs use the blocky JPEG compression but TGAs use another type of compression which is almost lossless which is why their file size is larger.

JPEG ≠ TGA

but every _SINGLE_ pixel is in the _EXACT_ same spot, and every _SINGLE_ pixel is the _EXACT_ same in each one. there are ABSOLUTELY no differences between the tga and jpg images. im talking about wolfcam, though, not QL. it doesnt matter what kind of compression wc uses for jpgs, because its obviously not noticeable. if it were noticeable, then i would have noticed it. if you save a jpg that is 1500k, and it uses a special type of compression that reduces the filesize by saving large sections as opposed to individual pixels, it will not be the same as a tga. you can save it as .bmp, and the filesize will increase for whatever reason, but the image itself will stay the same. each pixel in the avi, jpg, and tga was the exact same.

there is absolutely no difference whatsoever between tgas, avis and jpgs. they all have the exact same image data. neither one is lower quality than the other. tga shouldnt even be used anymore, because its a massive waste of space, and has less compatibility.

[fraggedICE]

ok so. i just capped 6 jpgs, 6 tgas and 6 avi frames. this was using /video, with cl_aviframerate 40 (not that it matters, because the demo was paused). I opened them all in photoshop, once again, and looked. I saw no differences. I saved each one as .bmp, because its lossless (the avi i flattened then saved, just fyi). I havent looked at the exact sizes, but each one is about the same size.

THE RESULTS:

final_avi.bmp - 4.94 MB (5,184,072 bytes) [original size: 22.2 MB (23,382,016 bytes)]
final_jpg.bmp - 4.94 MB (5,184,072 bytes) [original size: 10.6 MB (11,173,806 bytes)]
final_tga.bmp - 4.94 MB (5,184,072 bytes) [original size: 22.2 MB (23,328,108 bytes)]

the "original sizes" are from 6 frames, because the avi was 6 frames long. each jpg and tga were the exact same size, because it was the same scene. avis use the tga-style "compression", where each frame seems like its a "container" of a specific size, and no matter how much actual data is in the "container", it doesnt change size. so each frame is 3.70mb at 1440x900, whereas each jpg is 1.77mb at the same resolution.

just like i said, they are all the _EXACT_ same. i realized it while i was taking the screenshots, but you probably have some bad settings (chances are its r_jpegcompressionquality, default is 90, mine is at 100). This proves that jpg is the best POSSIBLE choice, in every way, aside from time saving, because you have to re-render as uncompressed .avi after. there might be other ways, but whatever. The size:quality ratio for jpgs is the best, no doubt.

You edited this post 1 min before I posted mine :X

But this doesn't help prove your point either, in fact it helps to prove that AVIs may actually be completely lossless, since you said the avi frame was 53 908 KB larger than the TGA one. Also just FYI r_jpegcompressionquality will not make the jpeg lossless, it's impossible to have a lossless JPEG (except for the JPEG-LS format lol, but I highly doubt it uses that).

And I never said the size:quality ratio of JPEGs wasn't the best, it is indeed the best, I was just proving that AVIs JPGs and TGAs are not equal.

but every _SINGLE_ pixel is in the _EXACT_ same spot, and every _SINGLE_ pixel is the _EXACT_ same in each one. there are ABSOLUTELY no differences between the tga and jpg images.

No. You just can't SEE the difference lol. I mean, none of this really matters at all and I do agree that JPEG should definitely be used over TGA... but anyway I did the same to my images, saved them as BMP, and compared them the same way as before and there was still the same amount of difference between JPEG.bmp + TGA.bmp as before. I'm sure if you lowered the r_jpegcompressionquality you would see a huge difference. Try putting it on 10 or something... every single pixel is not the exact same in each one. And yes I am referring to wolfcam in all of my posts.

There is absolutely no difference whatsoever between tgas, avis and jpgs. they all have the exact same image data. Neither one is lower quality than the other.

False, as proven by my experiment.

tga shouldnt even be used anymore, because its a massive waste of space, and has less compatibility.

I agree with this. I don't know why they use it exactly, but tga is mostly used in games for textures and many other games use it for saving screenshots (source from Wikipedia).

Edit: I can confirm that at r_jpegcompressionquality 100 it does not use JPEG-LS, and there is a difference between JPEGs at 100 quality and TGAs, although smaller than at 90 (there were hardly any fully white areas, but still colour offsets etc).

[Salamatiqus]

JPEG is lossy format and TGA is lossless. What's the problem?
I use tga. (And I listen .flac music)

[vanBasten]

l
NO DIFFERENCE BETWHEN "video jpg wav" AND "video videoname" I ALREADY TESTED THESE TWO FORMATS.
SO HERE ARE ANOTHER SUPPORTER FOR:
"JPG=TGA=AVI"

[KittenIgnition]

No. You just can't SEE the difference lol. I mean, none of this really matters at all and I do agree that JPEG should definitely be used over TGA... but anyway I did the same to my images, saved them as BMP, and compared them the same way as before and there was still the same amount of difference between JPEG.bmp + TGA.bmp as before. I'm sure if you lowered the r_jpegcompressionquality you would see a huge difference. Try putting it on 10 or something... every single pixel is not the exact same in each one. And yes I am referring to wolfcam in all of my posts.

if i cant see it, it doesnt matter at all. the only difference is the filesize. i pretty much looked at every section of each image, and there was no VISUAL difference. it doesnt matter how its done, because theyre the exact same image. if the definition of "lossless" isnt "no loss of image data", then its a pointless term, because theres no loss of image data in the jpgs when compared to the tgas. the filesize is different, because of compression etc., but no matter how closely you look or how you examine, jpgs are the same as tgas (visually).

http://www.mediafire.com/?kpnn4vrywtkpns7

thats a tga and a jpg, the exact same frame, with max jpg quality. if you can actually SEE a difference between each one, ill give you a cookie, because i cant see a single difference, and yet the jpg is less than half the size of the tga.

JPEG is lossy format and TGA is lossless. What's the problem?
I use tga. (And I listen .flac music)

if jpgs are the exact same as tga, but jpgs are lossy and tgas are lossless, then that word (lossless) really needs to be updated, because its literally wrong.

[KittenIgnition]

ok, so i sent the things to cyberstorm, and because hes such a dick he just had to prove me wrong. he downloaded some image comparer, and it turns out the images are only 99.9% similar or something. i looked at the images, and sure enough, if you look at 3200% zoom, you can see some differences, such as these: http://www.mediafire.com/?du2wop7l4hmnp3i

now, im no expert, but when you have to look at an image at 3200% zoomification, i think that its safe to say that theres no difference that actually matters. i cut out a section thats about 39 pixels wide, because thats as close as you have to be to notice a difference. the fact that a jpg can be compared with a lossless tga and have less than 1% difference, its pretty impressive. i guess its the style of compression or whatever, but the images are the same as far as im concerned. its not A=A=A kind of "the same", where theres 100% similarities, but over 99% is close enough, seeing as how a jpg is 1% different from a lossless tga, while being under 50% as big.

so, i was wrong, but i honestly cant say that these jpgs are lower quality than the tgas. this is 1440x900, uncompressed. if you put these clips in a video and compress it, it doesnt matter what codec you use, the end result will be the exact same. the difference is so minuscule you would have to have a mental disorder to actually care (but i cared enough to figure out that there was a difference at all, so i qualify as mentally challenged).

[fraggedICE]

ok, so i sent the things to cyberstorm, and because hes such a dick he just had to prove me wrong. he downloaded some image comparer, and it turns out the images are only 99.9% similar or something. i looked at the images, and sure enough, if you look at 3200% zoom, you can see some differences, such as these: http://www.mediafire.com/?du2wop7l4hmnp3i

now, im no expert, but when you have to look at an image at 3200% zoomification, i think that its safe to say that theres no difference that actually matters. i cut out a section thats about 39 pixels wide, because thats as close as you have to be to notice a difference. the fact that a jpg can be compared with a lossless tga and have less than 1% difference, its pretty impressive. i guess its the style of compression or whatever, but the images are the same as far as im concerned. its not A=A=A kind of "the same", where theres 100% similarities, but over 99% is close enough, seeing as how a jpg is 1% different from a lossless tga, while being under 50% as big.

so, i was wrong, but i honestly cant say that these jpgs are lower quality than the tgas. this is 1440x900, uncompressed. if you put these clips in a video and compress it, it doesnt matter what codec you use, the end result will be the exact same. the difference is so minuscule you would have to have a mental disorder to actually care (but i cared enough to figure out that there was a difference at all, so i qualify as mentally challenged).

Well, I also have a mental disorder... and I already decided to do a complete analysis of the screenshots xD

Download here.[1.4MB]

Like the other one, I have shown where the compression is (CompressionMap.png), here there are hardly any white parts as you used 100 quality for the compression. I edited it more though to show the white pixels (WhiteCompressionMap.png) and then I blurred/enhanced the white pixels to show the compressed areas better (BlurredMapOverlay.png). Behind the white blur is your original image, you can clearly see that most of the original image is only very dark/black parts, solid colour areas don't need any JPEG compression at all - the white parts are where there has been a lot of JPEG compression, although it seems like only very small changes, it does save a lot of disk space in the end.

I can upload the original [10MB] Photoshop document if you wish, I used smart filters and adjustment layers so you can change the amount of blur, remove some of the filters etc.

[KittenIgnition]

I can upload the original [10MB] Photoshop document if you wish, I used smart filters and adjustment layers so you can change the amount of blur, remove some of the filters etc.

id like that.

even though you found all that, all the compressed stuff is so minute that its completely invisible, even with a closeup look. you have to look REALLY hard to find the spots that were actually changed by the compression.

this went kind of off topic though XD

[fraggedICE]

id like that.

even though you found all that, all the compressed stuff is so minute that its completely invisible, even with a closeup look. you have to look REALLY hard to find the spots that were actually changed by the compression.

this went kind of off topic though XD

Here it is. [10MB]

So what can we draw from all of this? We now know there are no reasons whatsoever to use /video jpeg/tga over /video avi; avi has the highest quality, is most convenient etc. If you plan to take screenshots r_jpegcompressionquality should be set to 100 to minimize artifacts. It seems that the tga format is only used because it's a standard for most video games, and it provides lossless quality. This means we can also suggest that id provide an option to export jpegs rather than tgas with cg_autoaction 2/3, as it saves lots of space and there is hardly any quality difference. Also, if possible we could suggest to brugal to encode the avi files with Huffyuv to save disk space and to make extremely large clips compatible with video editors.

I've actually learned quite a lot thanks to this thread lol, even things like the difference between median/gaussian blur. I might just make a movie making guide myself with this info!

Edit: http://lags.leetcode.net/codec.html I just clicked this link in the OP, apparently this codec was based on Huffyuv, has better compression and can take advantage of multi core processors, sounds like another useful thing I shall be using from now on.