I feel like I'm on an endless quest. When I click a computer video, it refers me back to 3 other preface videos to watch.. and then those three have their own prefaces and it's just impossible to get back to the beginning
Is it possible to talk about the various kinds of anti aliasing in video games? The average "PC-channel" doesn't get very technical when taking about that. I would like that very much, as those are always the options to tinker with when hitting performance limits of your GPU
teekanne15 I found something interesting whilst messing out in the campaign of Battlefield 3 - everybody knows that Going Hunting is always used for benchmarks because its the most consistent in action, and it's also one of the low resource maps. On my 1920x1200 monitor, 4x MSAA is only very, very slightly better than 2x MSAA, but 2x MSAA is a WHOLE lot better than the OFF setting. But I fully agree
teekanne15 Yes, Supersampling > MSAA > FXAA (and others) But notice that this applies mostly to rasterization aliasing (geometry/triangle edges!), supersampling and FXAA can also fix texture aliasing but not enough. The Bilinear/Trilinear and Anisotropic settings you find in games help with the texture aliasing that happens due to resizing and skewing images to fit onto the model, and the GPU has dedicated hardware for that so they are relatively efficient to do and don't cost much :)
teekanne15 No, Anisotropic and Biliear/Trilinear sampling is separate from other Antialiasing techniques like MSAA, they solve different problems. MSAA is for triangle edges (the outlines of the 3d model), Bilinear sampling is for the inside, it prevents pixelation of the textures up until you try to scale it less than half, more than that you'll have to enable mip-maps (pre-calucalted high quality down scales) and preferably with Trilienar sampling to get a more accurate average value for the pixels. However if the texture is not facing the camera (angled) and not just scaled down then the far parts might get pixelated (or the near parts blurry), so Anisotropic takes even more samples when getting the average pixel value.
I actually learned this stuff in first year university in a catch all course called "Introduction to Multimedia Design". The chapter after this was about the history of Disney. The chapter before was about Copyright law. It was a fucking weird course.
JPG is still the most common image format for advertising on the web, and it's important to know how its quality relates to file size and thus page loading speed.
Oh no it's a cliff hanger. Now I've got to wait a week to find out what a dicrete cosine transform is. Great video. I'm looking forward to seeing the rest.
YCbCr *4:4:4* : no chrominance downsampling YCbCr *4:2:2* : horizontal chrominance downsampling by a factor of 2 YCbCr *4:2:0* : horizontal & vertical chrominance downsampling by a factor of 2 *SÉCAM* : for each frame, only store Cb or Cr, alternatively ITU-R *BT.2020* / ITU-R *BT.709* / ITU-R *BT.601* : what is "Red" / "Green" / "Blue" / "White"
Laura Halliday What he said here is SECAM alternates between Cb and Cr across frames (one frame is Y Cb, the next is Y Cr, and so on), how does that differ from what you said? and BTW the player I had supported both PAL and SECAM, this grayscale thing happened when I put a PAL tape and flipped the SECAM switch in the player.
Not sure if I'm alone here, but I find it easier to go through 1 half hour video than 3 10 minute ones. That said, the content on this channel is great.
I tried to explain lossless compression without knowing what it was called to a boss once by describing it as zipping a file. He thought it was impossible to get the same quality out as what was put in.
Not a great video... An excellent video with clear and concise explanations... Two parts to being an excellent teacher. Knowing every detail of the subject matter and the ability to express it in ways that others can absorb it... Excellent videos Michael. I can't thank you enough for breaking this down to a level that I can understand it... Thanks!
I make a lot of video captures to work with. I can create them in any file format. Which one should I use to make the best quality? Space is not a problem.
This sounds like there should be more options for how I want to compress an image when saving to jpeg than Photoshop gives us. There could in theory be a seperate value for the image compression and the color space compression, is that correct? So why isn’t Photoshop offering this option? Shouldn’t be to hard to implement. It’s like with 8-bit png where Photoshop doesn’t allow more than 1-bit alpha channel, even though it would be possible (Fireworks actually allowed this).
the jpeg format is actually the jpeg format because that is what the file extension is. It does not say anything about any internal container formats. Most file formats can actually contain different internal formats or compression algorithms, they are still that file format.
Watching this as a not very techy photographer is just having things I already understand explained to me by comparing them to things I know nothing about.
Cool ! I use JPEG so much, but I didn't know such details. Btw, my largest JPEG file is currently nasa hubble's carina_80x40full.jpg, 51 MB in size but 360 Megapixels, so probably a lossy compression of factor 7 or 8. But I can't notice the difference between TIFF and JPEG, same as any mp3 above 320 kb/s, it's impractical, sizewise, to have so much lossless detail, unless you have tremendous storage of several TB
i never encountered any software that could create an mp3 file with more than 320 kb/s is that even possible? i thought that was a limit of the mp3 standard.
4:58 so to sum up the entire vid, "corporate wants you to find the difference between those two pictures", for humans at normal screen distance, they are the same picture. Truly, a picture is worth 26,220 pictures! I assumed every second is 60 frames.
Greg Hornby You would save HALF the bandwidth if you downsample by that factor. In JPEG this is done for the Cr and Cb planes, meaning Y (Luminance) is left alone and Cr/Cb (Chrominance) is halved, resulting in a 1/3rd bandwidth saving. No downsampling one pixel would be [Y, Cr, Cb] = 3 bytes per pixel With 4:2:0 downsampling, one Cr/Cb now covers 4 Y pixels, so here it gets a bit trickier to compute... [Y,Y,Y,Y, Cr,Cb] = 6 / 4 = 1.5 bytes per pixel. 4:2:2 would be Y,Y,Cr,Cb = 4/2 = 2 bytes per pixel.
I remember when I want to convert my xmind file into a picture... First, I did a BMP file, which was about 200MB. The size of it is terrible, so I did a JPG file, which was KBs. However, when I zoom out, I find it's impossible to recognize the word on the map. Finally, I had to do a PNG file, about hundreds of KBs. So in my opinion, jpg file is worth doing when the exact contents are not that necessary when the picture is big.
digdugdiggy I think it's not only about storage, but also about transfer. The bandwidth for mobile phone internet for example is usually not as high as the one you get at home, so it'd be unwise to send these huge files over your mobile phone. There's probably also a lot of legacy involved.
digdugdiggy On a small scale, like keeping a handful of photos on your PC, sure, you might as well keep the image quality. On a large scale, like a server hosting hundreds of thousands, or possibly millions of photos, you'll want to have the most efficient use of your space as possible. Although I could have all of my photos be lossless, I value storage space more than I do a couple pixels being "more correct". There are much more important things I could have on my computer. Like lossless music.
I hope there's a mention of JPEG 2000. 20% more efficient and less prone to artifacts at lower resolutions, it's a wavelet based compression (as opposed to JFIF's discreet cosine transform). For a while it was touted as the successor to the JPEG (ie. JFIF) format. Even though it's supported by many software applications you almost never find it "in the wild".
MrSlowestD16 I should have been clearer. I meant that while it's supported by some major image processing programs like Photoshop (although in that case you have to manually install it from the disk), most viewers and browsers can't handle it.
SerBallister There were patents involved but they may have been pre-existing. It was the poorly implemented licensing plan in an already glutted market that did it in.
I used to hate JPG. Back when I was using Microsoft Paint (for Windows Vista) to draw everything, MS Paint would save stuff as JPG as default, and ruin my drawings with noise. Now I know the difference, but I still hesitate over the compression level when saving as JPG in Photoshop.
the problem in your case isn't just the JPG format itself, but that Microsoft Paint did a very bad job at making those. (or any format other than BMP, even messed up GIF pretty bad by forcing everything into one standard palette using dithering; messed up PNG by making them huge ever for PNG)
DCT is hardly a "complicated mathematical method". It's a fairly common and relatively straightforward mathematical method for physicists and engineers.
***** I disagree. It's not "impossible to understand for somebody who isn't an engineer or physicist." I think as a computer science graduate I technically don't fall into either category I don't think, it's just a title. I can sit down and teach somebody who works at mcdonalds if I wanted. Don't pretend human capabilities are limited by their degree or lack there-of. It's just a piece of paper. The only limiting factor is peoples' will to put forth the effort.
Hehe, this discussion is kinda related to the topic of JPEG compression, in a very abstract way... discarding high frequency information because it doesn't matter much for the big picture.
mindrapeart How dare you! Anathema! :-D But seriously, TIFF files are notoriously difficult to exchange between different systems, as they might use different encoders and decoders, and TIFF is really more of a container format, rather than a compression algorithm. The only other lossless format that doesn't have the TIFF issues is webp, which if it works at all will work the same everywhere. It can be both lossless or lossy, depending on what's needed.
Do you mean for the web? I use 16 bits per channel when image making for print, not 8. I also use layers. Png is useless for saving in 16 bit with layers. Pretty sure i'm right. Please feel free to fill me in if i'm not. Thanks.
mindrapeart IMHO, if you really need layers, it means you also need to manipulate the image, and at that point, you may as well store it in the editor's format, such as .psd for example, which I think also supports 16 bit channels. It's not highly portable between different editors, but then again, neither is TIFF. Due to the ubiquity of Photoshop, there are more psd converters (to other editor specific formats) out there then converters for different TIFF implementations. If you need to actually show the image, then you don't need the format to support layers, and you only need to have it store enough colors to accommodate the environment. In the case of screens (whether it be web, or a presentation, a photo on a phone or whatever), 8 bits per channel is what screens are using anyway, so you don't need anything more. For print, sure, 16 bits is nice to have. The only format to support 16 bit channels that is lossless and has alpha transparency (=> everything that PNG is, and more) is JPEG XR (or ".jxr"). But the only browser that supports it is IE9 and above, and most editors still need plug-ins for it, if there are any to begin with. Until such support is more common, PNG is, at least for screens, the better format to use.
mindrapeart PNG does indeed not support layers, but it does support 16 bits per channel. A more pressing issue when using PNG for print is that it does not support CMYK and will always store colours in full RGB or an RGB-derived palette, but then if I were printing an image, I'd just turn it into a PDF or keep it as PSD, if the printer could handle PSD files. For use on the internet though, PNG is by far the best image file format, in my opinion.
why do you have breakfast running all the time, it wastes damn much power, or creates tons of heat, whatever, have it pulse served, like it should, processors
no I dont mean cores turned off, the logic units, when not used, if you can switch logic units to other purposes, turned off, not sending anything, zeros or ones, when not used, like breakfast facilities, you are not in the facilities all the time, either, duh
blizzy1978 pronouncing PNG as “ping” is just like saying “gif” with a hard “g”. Sure, it is “correct”, but it sounds absolutely awful and no one should pronounce it like that. Ever.
blizzy1978 That’s pretty much the reason I put “correct” in quotation marks. But, when it comes to language, anything accepted by a large number of people is considered correct. The majority of people (afaik) say “gif” with a hard G, therefore it can be considered correct. So both are correct, but the soft G version is obviously the best one.
Walt Mays In text, if I write JIF, people will immediately know what I’m referring to. In an actual conversation, I would probably just say “JPEG Interchange Format” to avoid confusion. Thing is, I never needed to talk to anyone about JIF, and I don’t think I ever will. And, really, if I say GIF with a J sound, I *really* doubt anyone will think I mean JIF. In other words, except for a few very specific contexts, it is a distinction that doesn’t need to be made in the first place.
Walt Mays Better yet: how do you make the distinction between “then” and “than”, “affect” and “effect”, “your” and “you’re”, “its” and “it’s”, “queue” and “cue”, singular “you” and plural ”you”?
***** Why he's definitely not a moron: JPEG destroys pixel art. One misplaced click can ruin hours of work in under a second. He's a moron for thinking jpeg looks terrible? Tell that to every designer who's ever been fired for accidentally clicking "save" instead of "save as" once.
notoriouswhitemoth use a backup system with versioning. Overwriting a file isn't as big a deal anymore, you can just restore the last good version. Also waiting hours in between saves isn't a good idea.
notoriouswhitemoth This is why version control exists and it's not only for software development. Or if you wanna skip all the complexity, just use Dropbox.
JPEG is a file format. Anything after a '.' character the computer automatically turns it into that type of file format. You can name a file name.computerfile and it will become a COMPUTERFILE File. While the naming of the file format being JPEG was probably not intended, that is what is has become. Just because it was supposed to be known as something else doesn't mean anything other than history.
Pet peeve: Times less. "There will be 100 times less color" If there was 1 times less color, there would be zero color, WTF does 100 X less color look like? It makes no sense. Did you mean 1% as much color?
Hahaha I love the dramatic zoom put in when he says "however" at 3:19
+Jombo cameraman slipped or fell asleep.
Very Tarantinoesque
That's so odd. I glanced down and read your comment at 3:17 and then it happened. Brilliant timing, well done.
the Office irl
I feel like I'm on an endless quest. When I click a computer video, it refers me back to 3 other preface videos to watch.. and then those three have their own prefaces and it's just impossible to get back to the beginning
Oh nice, there are three more videos (to the power of three (to the power of three (to the power of three (...))))
Hence why I just started watching them all from the beginning.
@@cimmik to watch computerphile first you must watch computerphile
@@cimmik literally tail recursion lol
ok?
Is it possible to talk about the various kinds of anti aliasing in video games? The average "PC-channel" doesn't get very technical when taking about that. I would like that very much, as those are always the options to tinker with when hitting performance limits of your GPU
teekanne15 Seconded!
teekanne15 I found something interesting whilst messing out in the campaign of Battlefield 3 - everybody knows that Going Hunting is always used for benchmarks because its the most consistent in action, and it's also one of the low resource maps. On my 1920x1200 monitor, 4x MSAA is only very, very slightly better than 2x MSAA, but 2x MSAA is a WHOLE lot better than the OFF setting. But I fully agree
teekanne15 Yes, Supersampling > MSAA > FXAA (and others)
But notice that this applies mostly to rasterization aliasing (geometry/triangle edges!), supersampling and FXAA can also fix texture aliasing but not enough. The Bilinear/Trilinear and Anisotropic settings you find in games help with the texture aliasing that happens due to resizing and skewing images to fit onto the model, and the GPU has dedicated hardware for that so they are relatively efficient to do and don't cost much :)
Hassan Selim
So Anisotropic filtering only affects the visual fidelity when applying a form of AA before ?
teekanne15 No, Anisotropic and Biliear/Trilinear sampling is separate from other Antialiasing techniques like MSAA, they solve different problems.
MSAA is for triangle edges (the outlines of the 3d model), Bilinear sampling is for the inside, it prevents pixelation of the textures up until you try to scale it less than half, more than that you'll have to enable mip-maps (pre-calucalted high quality down scales) and preferably with Trilienar sampling to get a more accurate average value for the pixels. However if the texture is not facing the camera (angled) and not just scaled down then the far parts might get pixelated (or the near parts blurry), so Anisotropic takes even more samples when getting the average pixel value.
what a brilliant educator - providing us with superbly understandable descriptions of these complex matters. thank you greatly.
Very clear explanation! Before this video, it was hard for me to understand image compression basics. You've helped me a lot. Thank you very much.
I actually learned this stuff in first year university in a catch all course called "Introduction to Multimedia Design". The chapter after this was about the history of Disney. The chapter before was about Copyright law. It was a fucking weird course.
hmmm, that (JPEG compression) seems to me as a weird choice for something to learn in a course of that name...
JPG is still the most common image format for advertising on the web, and it's important to know how its quality relates to file size and thus page loading speed.
ok?
2:44 BMP is (almost always) not compressed. It is raw data, hence why those files are huge.
Oh no it's a cliff hanger. Now I've got to wait a week to find out what a dicrete cosine transform is.
Great video. I'm looking forward to seeing the rest.
As a photographer, I found this video to be very useful.
I really hope part 2 is coming soon. I NEED my Discrete Cosine Transforms!
Photographers and needlessly overcomplicating things, name a more iconic duo.
This is great for me because I am doing a project that focuses on JPEG Steganography and this video was very helpful.
Can't wait to see the rest!
YCbCr *4:4:4* : no chrominance downsampling
YCbCr *4:2:2* : horizontal chrominance downsampling by a factor of 2
YCbCr *4:2:0* : horizontal & vertical chrominance downsampling by a factor of 2
*SÉCAM* : for each frame, only store Cb or Cr, alternatively
ITU-R *BT.2020* / ITU-R *BT.709* / ITU-R *BT.601* : what is "Red" / "Green" / "Blue" / "White"
Son-Tchor I Now I know why when I pressed the secam button on my VHS player for a PAL signal the image became grayscale! Thanks for the info :)
Hassan Selim
Actually, it's because SECAM video tapes record colour informaiton in a way that PAL players can't read.
Laura Halliday What he said here is SECAM alternates between Cb and Cr across frames (one frame is Y Cb, the next is Y Cr, and so on), how does that differ from what you said?
and BTW the player I had supported both PAL and SECAM, this grayscale thing happened when I put a PAL tape and flipped the SECAM switch in the player.
Hassan Selim That's interesting, so the colour planes update at half the frame rate ?
SerBallister That's what I understood from the original comment here, I didn't read about this before :D
I love the number of times "There was quite a lot, so no one did it" or similar shows up.
Mike Pound is a hero!
I actually used colorspace conversion in an app I'm making because of these videos. Nice work guys, keep it up! :)
2:45 BMP is raw data though (ie. Bit MaP), it's not a compression method.
scbtripwire BMP supports optional run-length encoding.
Great videos. Have you got any video showing the jpeg file format? Thanks
Not sure if I'm alone here, but I find it easier to go through 1 half hour video than 3 10 minute ones. That said, the content on this channel is great.
I tried to explain lossless compression without knowing what it was called to a boss once by describing it as zipping a file. He thought it was impossible to get the same quality out as what was put in.
I heard about .bpg recently and I really want to know how this works?
What is 'downsample'? Does downsampling by 2 mean that we take an average value of 2x2 pixel blocks?
Not a great video... An excellent video with clear and concise explanations... Two parts to being an excellent teacher. Knowing every detail of the subject matter and the ability to express it in ways that others can absorb it... Excellent videos Michael. I can't thank you enough for breaking this down to a level that I can understand it... Thanks!
at 4:40, "human eye doesn't really see ____ very well" ........
what's the exact word in the gap ? i can't really get it.
chrominance
Did lot of Lossy and Lossless compresion stuff in Computing in High School.
Only touched on Jpeg formats though.
It would have been good if the picture with downsampled luminance was shown, to better illustrate the difference in color/brightness perception.
I wonder if there are any online interactive demonstrations of these principles
Why only use cosines? What about any other general smooth bounded basis function?
question then why is jpeg a file type on my computer
I make a lot of video captures to work with. I can create them in any file format. Which one should I use to make the best quality? Space is not a problem.
TIFF is lossless compression using LZW, but for the reasons stated in my previous post I hold that PNG is a better lossless format.
This sounds like there should be more options for how I want to compress an image when saving to jpeg than Photoshop gives us. There could in theory be a seperate value for the image compression and the color space compression, is that correct? So why isn’t Photoshop offering this option? Shouldn’t be to hard to implement.
It’s like with 8-bit png where Photoshop doesn’t allow more than 1-bit alpha channel, even though it would be possible (Fireworks actually allowed this).
Great stuff! Really well communicated.
what if the original imag is a greyscale picture rather a RGB picture?
the jpeg format is actually the jpeg format because that is what the file extension is. It does not say anything about any internal container formats. Most file formats can actually contain different internal formats or compression algorithms, they are still that file format.
Watching this as a not very techy photographer is just having things I already understand explained to me by comparing them to things I know nothing about.
I'm in agony, waiting to find out how DCTs work. Any chance of a post soon? It's been a week :)
Cool ! I use JPEG so much, but I didn't know such details. Btw, my largest JPEG file is currently nasa hubble's carina_80x40full.jpg, 51 MB in size but 360 Megapixels, so probably a lossy compression of factor 7 or 8. But I can't notice the difference between TIFF and JPEG, same as any mp3 above 320 kb/s, it's impractical, sizewise, to have so much lossless detail, unless you have tremendous storage of several TB
i never encountered any software that could create an mp3 file with more than 320 kb/s
is that even possible? i thought that was a limit of the mp3 standard.
can you make a video about "trellis quantization" in JPEG?
4:58 so to sum up the entire vid, "corporate wants you to find the difference between those two pictures", for humans at normal screen distance, they are the same picture.
Truly, a picture is worth 26,220 pictures!
I assumed every second is 60 frames.
"jpeg is not a file format" im just 10 seconds in and my mind is already blown
So if you downsample by a factor of 2 only vertically, you'd save 1 bit for every pixel?
Greg Hornby If the colour values were stored as 8-bit like he suggests then you would save 16 bits for every pixel: 8 from the Cb and 8 from Cr.
Greg Hornby You would save HALF the bandwidth if you downsample by that factor. In JPEG this is done for the Cr and Cb planes, meaning Y (Luminance) is left alone and Cr/Cb (Chrominance) is halved, resulting in a 1/3rd bandwidth saving.
No downsampling one pixel would be [Y, Cr, Cb] = 3 bytes per pixel
With 4:2:0 downsampling, one Cr/Cb now covers 4 Y pixels, so here it gets a bit trickier to compute... [Y,Y,Y,Y, Cr,Cb] = 6 / 4 = 1.5 bytes per pixel.
4:2:2 would be Y,Y,Cr,Cb = 4/2 = 2 bytes per pixel.
Ok, downsampling I do know. I've never understood DCT. Please upload soon that video.
I remember when I want to convert my xmind file into a picture... First, I did a BMP file, which was about 200MB. The size of it is terrible, so I did a JPG file, which was KBs. However, when I zoom out, I find it's impossible to recognize the word on the map. Finally, I had to do a PNG file, about hundreds of KBs.
So in my opinion, jpg file is worth doing when the exact contents are not that necessary when the picture is big.
Why are we worried about compression still when storage space is so plentiful? Why not keep all the image quality?
digdugdiggy I think it's not only about storage, but also about transfer. The bandwidth for mobile phone internet for example is usually not as high as the one you get at home, so it'd be unwise to send these huge files over your mobile phone. There's probably also a lot of legacy involved.
digdugdiggy On a small scale, like keeping a handful of photos on your PC, sure, you might as well keep the image quality. On a large scale, like a server hosting hundreds of thousands, or possibly millions of photos, you'll want to have the most efficient use of your space as possible.
Although I could have all of my photos be lossless, I value storage space more than I do a couple pixels being "more correct". There are much more important things I could have on my computer. Like lossless music.
Awesome! I am looking forward to the next videos!
I am seriously into these videos. This is beyond interesting. thank you for these!
Do I look like I know what a JPEG is?
Looks like part of longer series on digital imaging to me.
is there a jpeg file parts documentation?
I hope there's a mention of JPEG 2000. 20% more efficient and less prone to artifacts at lower resolutions, it's a wavelet based compression (as opposed to JFIF's discreet cosine transform). For a while it was touted as the successor to the JPEG (ie. JFIF) format. Even though it's supported by many software applications you almost never find it "in the wild".
MrSlowestD16 I should have been clearer. I meant that while it's supported by some major image processing programs like Photoshop (although in that case you have to manually install it from the disk), most viewers and browsers can't handle it.
RMoribayashi Gotcha, I agree 100%
RMoribayashi Didn't they get greedy and patent parts of the algorithm, killing its uptake ?
SerBallister There were patents involved but they may have been pre-existing. It was the poorly implemented licensing plan in an already glutted market that did it in.
floatboth WebP is DCT based when in lossy mode ?
You could take the unnecessary 'u' out of color and that would reduce the size by 1/6 right there.
Nice vídeo 👏
cross your eyes at 5:20 and get a 3d flower
I used to hate JPG. Back when I was using Microsoft Paint (for Windows Vista) to draw everything, MS Paint would save stuff as JPG as default, and ruin my drawings with noise.
Now I know the difference, but I still hesitate over the compression level when saving as JPG in Photoshop.
the problem in your case isn't just the JPG format itself, but that Microsoft Paint did a very bad job at making those. (or any format other than BMP, even messed up GIF pretty bad by forcing everything into one standard palette using dithering; messed up PNG by making them huge ever for PNG)
I like PNG images as they are lossless (even though they take up more space).
PNG has also alpha channel, allowing transparency
PNG is common, but BPG is better (as long as x265 is free. WebM's VP9 is safe.), and FLIF is even better (if you don't need audio).
I recall being oblivious to the purpose of PNG and using bloated BMP files for too long - as editing masters.
does someone can add english subtitles please ?
Am I the only one who could immediately tell the difference between the two images and knew the one on the right was the compressed one instantly?
what is a high frequency change ?
continue watching the series, it will be explained
In the title colour is spelled the correct Canadian way with a “u”
ahhhh my head, all those acronymes. I don't understand anything
This guy looks like he knows what a JPEG is.
This guy would be an excellent professor! or is he already ?
thank you! great video
Do I look like I know what a jpeg is?
b3l14l I was just thinking about that
100 or 10? 5:25
Funny how it only now occurs to me if some aliens saw our photographs they might look incredibly strange and cryptic compared to the subjects.
First rule of the jpeg is do not talk about jpeg
Jpeg was invented in 1984 at Rennes Atalante for the Minitel.
I guess linking that to MPEG makes sense then
Mind blowing
Pls add subtitles as well. British accent difficult to understand.
cutting when the best part starts... cliffhanger
Would be nice to see a video on YCbCr and YCoCg etc...
DCT is hardly a "complicated mathematical method". It's a fairly common and relatively straightforward mathematical method for physicists and engineers.
***** Then don't alineate them by pretending it's something impossible to understand.
***** I disagree. It's not "impossible to understand for somebody who isn't an engineer or physicist."
I think as a computer science graduate I technically don't fall into either category I don't think, it's just a title.
I can sit down and teach somebody who works at mcdonalds if I wanted. Don't pretend human capabilities are limited by their degree or lack there-of. It's just a piece of paper. The only limiting factor is peoples' will to put forth the effort.
Kilgore Trout Most computer programmers I know do not understand what DCT is. Unless you are specifically working on video codecs.
Hehe, this discussion is kinda related to the topic of JPEG compression, in a very abstract way...
discarding high frequency information because it doesn't matter much for the big picture.
well, technically BMP is not a compression.
SpackoEntertainment Optional RLE.
SpackoEntertainment may also reduce the palette
yes.............
Do ah look loik ah know wot a jay peg is?
Didn't know Tobey Maguire was an expert on Jpeg
editing error at 6:44, great video otherwise!
Thanks
나 이거 왜...?
Computer Graphics preventing me from getting my Masters Degree lol.
Why have loss if you don't have to? PNG ftw!
Robert Faber Or TIFF
mindrapeart How dare you! Anathema! :-D
But seriously, TIFF files are notoriously difficult to exchange between different systems, as they might use different encoders and decoders, and TIFF is really more of a container format, rather than a compression algorithm.
The only other lossless format that doesn't have the TIFF issues is webp, which if it works at all will work the same everywhere. It can be both lossless or lossy, depending on what's needed.
Do you mean for the web? I use 16 bits per channel when image making for print, not 8. I also use layers. Png is useless for saving in 16 bit with layers. Pretty sure i'm right. Please feel free to fill me in if i'm not. Thanks.
mindrapeart IMHO, if you really need layers, it means you also need to manipulate the image, and at that point, you may as well store it in the editor's format, such as .psd for example, which I think also supports 16 bit channels. It's not highly portable between different editors, but then again, neither is TIFF. Due to the ubiquity of Photoshop, there are more psd converters (to other editor specific formats) out there then converters for different TIFF implementations.
If you need to actually show the image, then you don't need the format to support layers, and you only need to have it store enough colors to accommodate the environment.
In the case of screens (whether it be web, or a presentation, a photo on a phone or whatever), 8 bits per channel is what screens are using anyway, so you don't need anything more. For print, sure, 16 bits is nice to have.
The only format to support 16 bit channels that is lossless and has alpha transparency (=> everything that PNG is, and more) is JPEG XR (or ".jxr"). But the only browser that supports it is IE9 and above, and most editors still need plug-ins for it, if there are any to begin with. Until such support is more common, PNG is, at least for screens, the better format to use.
mindrapeart PNG does indeed not support layers, but it does support 16 bits per channel. A more pressing issue when using PNG for print is that it does not support CMYK and will always store colours in full RGB or an RGB-derived palette, but then if I were printing an image, I'd just turn it into a PDF or keep it as PSD, if the printer could handle PSD files. For use on the internet though, PNG is by far the best image file format, in my opinion.
Awesome
jpeg joint photographic experts group
*color
why do you have breakfast running all the time, it wastes damn much power, or creates tons of heat, whatever, have it pulse served, like it should, processors
no I dont mean cores turned off, the logic units, when not used, if you can switch logic units to other purposes, turned off, not sending anything, zeros or ones, when not used, like breakfast facilities, you are not in the facilities all the time, either, duh
turn of the stove, when not used
interesting
K
joint photographic experts group
or what xd
lol it's right
I hate jpeg.
Long live PNG! We have TB HDDs, plenty of space!
MinecraftEpicPlayer Tell that to a smartphone user
not trying to be too picky, but "PNG" is pronounced "ping"
blizzy1978 pronouncing PNG as “ping” is just like saying “gif” with a hard “g”.
Sure, it is “correct”, but it sounds absolutely awful and no one should pronounce it like that. Ever.
blizzy1978 That’s pretty much the reason I put “correct” in quotation marks.
But, when it comes to language, anything accepted by a large number of people is considered correct. The majority of people (afaik) say “gif” with a hard G, therefore it can be considered correct.
So both are correct, but the soft G version is obviously the best one.
Walt Mays
In text, if I write JIF, people will immediately know what I’m referring to.
In an actual conversation, I would probably just say “JPEG Interchange Format” to avoid confusion. Thing is, I never needed to talk to anyone about JIF, and I don’t think I ever will.
And, really, if I say GIF with a J sound, I *really* doubt anyone will think I mean JIF.
In other words, except for a few very specific contexts, it is a distinction that doesn’t need to be made in the first place.
Walt Mays
Better yet: how do you make the distinction between “then” and “than”, “affect” and “effect”, “your” and “you’re”, “its” and “it’s”, “queue” and “cue”, singular “you” and plural ”you”?
***** “then” and “than”, “affect” and “effect” should all sound different to you. If they don't, you're not paying enough attention to your vowels.
You mean Jay feg, if you follow the logic of 'hard g gif' heathens :^)
how jpeg works: it makes everything look like dogshit
TekkGnostic Well, obviously you shouldn't use JPEG for text. It's great for photographs though.
***** Why he's definitely not a moron: JPEG destroys pixel art. One misplaced click can ruin hours of work in under a second. He's a moron for thinking jpeg looks terrible? Tell that to every designer who's ever been fired for accidentally clicking "save" instead of "save as" once.
notoriouswhitemoth use a backup system with versioning. Overwriting a file isn't as big a deal anymore, you can just restore the last good version. Also waiting hours in between saves isn't a good idea.
notoriouswhitemoth Why are you bringing back my horrible memories
notoriouswhitemoth This is why version control exists and it's not only for software development. Or if you wanna skip all the complexity, just use Dropbox.
JPEG is a file format. Anything after a '.' character the computer automatically turns it into that type of file format. You can name a file name.computerfile and it will become a COMPUTERFILE File. While the naming of the file format being JPEG was probably not intended, that is what is has become. Just because it was supposed to be known as something else doesn't mean anything other than history.
Khaltazar that makes it into an extension not a format
He looks like Peter Parker :D
With Snowdenesque traits as well. High enough JPEG compression would blur them out.
I love the subject of your videos but your way to speak is hard to catch for me .-.
omg... this british accent....
Pet peeve: Times less. "There will be 100 times less color"
If there was 1 times less color, there would be zero color, WTF does 100 X less color look like? It makes no sense.
Did you mean 1% as much color?
Jpeg is my nickname in a few games... watching this video is so fucking weird