i read almost every post that lands in the VSL forum. except the ones with stupid subjects, or asking questions that have already been answered. yours looked interesting so i thought i'd get in here and say something.

oh and thanks. glad the tutorial helped you out. that's why i put it there. hopefully more people will check it out. it probably answers a lot of questions the new video editors are wondering about.

to answer your post, the ProRes422 will not cause any generation loss. many folks have compressed and recompressed ProRes422 up to like 20 times and not noticed a difference visually. imagine compressing a file and taking the compressed file and compressing that. repeat 20 times. thats some damn good compression if it still looks good. that's prores.

you will suffer quality loss but it's because of the mp4 -> mp4 re-encoding.

put more simply, the:

mp4 -> ProRes422 - mp4

will not look any worse than:

mp4 -> mp4

Never use DV as an intermediate codec though. Your video will look horrible. DVs biggest drawbacks are the color subsampling (4:1:1) and extreme 5:1 compression, which is intraframe (each frame is encoded separate). Intra is very nice and edit friendly but nearly any advanced codec using any length GOP will beat DV easily is easily noticeable picture quality.

The poor 4:1:1 subsampling, along with 8bit precision (compared to ProRes422 HQ 10bit) will result in banding in your gradient areas, fades and so on. This gets even worse if your sequence is using DV too, so your transitions and fades, effects, titles, etc are probably being rendered with 8bit 4:1:1 precision too, depending on how you render your project.

In short, stay away from DV like the plague. Use ProRes422 as your intermediary if you're on Mac. Optionally use re-connect to media if using ProRes422. Use Cineform as your intermediary if you're on a PC (also 10bit 4:2:2, but may be a Wavelet based compression scheme, don't remember). Get the original VOB source if you can and stay away from re-editing mp4 files (if possible) since a LOT of the original picture data has been thrown away. If you must use DV as an intermediary pleas be sure and use re-connect media before rendering your final mp4, with the caveat about a DV sequence causing your titles, overlays, gradients, efx and transitions to look like crap, depending on how you render.

swift is my hero

Hehehe, well A-Swift, I've read your last post about four times and still don't understand it all, but at least after the first read I get the point. What more can I say: thank you!

Yeah, the FCP sessions should deserve a sticky or a link on the serato website itself. 100% "need to know" if you're getting into the VSL thing...

Greetz,
NLS

yeah, why don't these forums mods sticky my FCP Sessions!

b/c you're not Konix.

yeah a-swift,...you're always a huge help. i've learned a lot from you. props bro!

Well I'll break it down more simply. The color subsampling is a big deal so I'll focus on that. Without sending you to the wiki page, what does 4:4:4, 4:2:2 and 4:1:1 (DV) mean. Why is 4:4:4 awesome and why does 4:1:1 suck?

Well let's look at 4:4:4 subsampling in RGB color space. In a really simplified explanation, 4:4:4 means you have a bunch of pixels and the exact value for each pixel is stored uncompressed. This results in really excellent image but uses a ridiculous amount of storage space. For reference, a 720p 4:4:4 RGB video would use about TEN GB PER MINUTE. That is not a typo. A music video would be like 40GB. Obviously 4:4:4 RGB is not something we can work with. It's too much data. Bump that up to full HD and it's even more data.

So what does 4:2:2 mean. Well, in this example we'll use 4:2:2 with Y'Cb'Cr' colorspace. This is a different color space, but that just means we're using a different way for the pixels to represent the picture. Like RGB, this space is made by splitting the video signal into 3 components. Y represents the LUMA component (brightness). The other two components represent CHROMA (color) values. The third color can be calculated on the fly to reconstruct the full 3 color info but most of that is not important or probably technically accurate, but you get the idea. Here's what is important. With 4:2:2 subsampling we dedicate a full bunch of bits to the Y LUMA component. That's what the first 4: means. Then for the next components (the two chroma components), we throw away HALF the values! That's how 4:2:2 subsampling works. It works by throwing away HALF the values for the two chroma components.

4:2:2 subsampling uses 2/3 the storage of 4:4:4 and is visually lossless. When using a good codec with 10bit precision, and 4:2:2 subsampling, this can be very good.

Now DV on the other hand, starts with a 4:1:1 subsampling scheme so it's gonna be really tough to get good colors with that. Remember in 4:2:2 we threw away HALF the color information from our original picture data, to save 2/3 storage. Now take that color info and throw away HALF of those. That's 4:1:1 subsampling.

For guys that really know this stuff, my simple explanation may seem a little too simple but I wanted to really point out what's important, and that's that a 4:1:1 codec and the extreme loss of color info make it a REALLY horrible choice for an intermediate codec.

This is important guys, even if you don't want to understand what all the stuff I wrote above, read the above sentence again and just say no to crappy codecs that will kill your quality.

Video "crapware" spammers are fucking relentless.

Like mosquitos.

**slap**