Why isn’t everybody uploading H.264 codec?

Jason, what's your problem to say "I need more than help"? Surely we become no friends after this stupid comment, so it's better to stop this topic. It's really no help if you write as if i'm stupid, so better forget it!

Hi,

I thought I'd float this topic to the top again. I haven't processed for a while and thought I would do some tests before I started exporting. There were good points made in favour of photo-jpeg so I thought I would try several clips exported in the h.264 and photo-jpeg and ProRes422. I exported all of them in .mov using Adobe Media Encoder with files prepped in PrPro. I didn't make any changes in AME except the codec selected. I used about 95% quality. The p-jpeg output shows twice the bit rate and twice the size as the h-264 but viewing it, the p-JPEG has noticeably more artifacts and banding, much more so than the h.264. The ProRes files are even bigger than the p-jpeg and to me, only look about the same as the h.264. If buyers prefer photo-jpegs, I don't mind delivering them, but I can't seem to make them look as good as the h.264. Any suggestions?

I am on h264 MOV Wagon.

Zany ya beat me to it!

I was testing the PhotoJpeg myth myself this week.

- Shot some tripod-ed, panned blue sky with my Canon 5DM2 which has native H.264 codec. Blue sky will show artifacts and banding very well.

- Exported untouched native file in MPEG Streamclip to PhotoJpeg at 90% quality, non-interlaced, unscaled.

Results: Extreme banding and artifacts and much larger file size. The trimmed export of the original H.264 file was MUCH cleaner than the PhotoJpeg. And a smaller file size.

I am new to video post so perhaps I am missing something because I find it hard to believe that I just discovered what an inferior codec PhotoJpeg is.

Anyone else want to try the “blue sky test” with a DSLR?

"" Anyone else want to try the “blue sky test” with a DSLR? ""

No God,oh please dont make me do it, nooooooooo. :)
Joking aside, to be able to have decent gradients on any kind of 8-bit digital footage,let alone a dslr,you need to expose for the sky to the right as much as you can,considering that you will be able to pull it down to a decent level (but not too much) .
That is if you plan to grade heavily.
If not, expose normally and leave the sky out of heavy hue changes while grading (for example, PLEASE, no secondaries)

Photojpeg is a standard editing codec,that can ingest in every nle available.If it does that much damage to a seemingly ok native h264 clip, then maybe one has to check whether there is any gamma change in the exported version or general export options on one's nle that are not optimum.Other that that it should be fine for most occasions.

I admit that I started outputting my final submission clips on H.264 to save file space thinking that if I'm shooting in H.264, editing in H.264, I might as well output in H.264 But after all the discussion on sales being hindered because photo-JPEG is the industry standard etc., I had to rethink my approach. I did a few tests and none of them on blue sky. But banding or blocky artifacts are present everywhere with photo-JPEG and this in a file that is twice as big to store. The photo-jpeg version has a grainier feel on playback. ProRes422 has about the same look in quality as H.264 but a file size that is about 2.5 times as big.
On the sample below the bottom of the screen grab is H.264 then photojpeg then ProRes422

http://i1277.photobucket.com/albums/y490/setsail2/ScreenShot2014-10-21at12957PM_zps32204df1.png

@BigTree a couple of years back I've also used Mpeg Streamclip but the output was not great but at that time I didn't know why. Yes I am pixelpeeping to get the best quality possible.

If you do the same test with the PRO workflow in AfterEffects CS6 or newer that I posted a couple of pages back you will get PRO results that are broadcast quality. Using a very old freeware tool will not get you there. So your test is wrong because Mpeg Streamclip does not a good job like AfterEffects when transcoding to a higher color space like 422.

The output from Mpeg Streamclip can't compete with the 32Bit floating point internal engine from After Effects. Since CS6 it is superior. 2 years ago I've talked also to the Cineform guys about that and they confirmed my tests too.

Here is my blog post about CS6 engine:
http://schweinert.com/blog/files/9d53ae3388c268d19ce9052989cf9a9a-15.html

PS. If you still want to upload H264 footage just do it. PRO's will always use edit friendly codecs whenever possible. But you have also to transcode it "correctly" if you have 8 Bit source material or the whole PhotoJPEG workflow is completely pointless and you will only get huge filesizes without added benefits. Just my two cents.

Same applies to cheaper software products like Magix, PowerDirector, ... they do not have the same transcode quality like AfterEffects. You can really say you get what you pay for. Except for DaVinci Resolve :-) but it can't output Photo JPEG anyway.

@danielschweinert FWIW I use the latest Creative Cloud software from Adobe to edit and transcode. I'm all for supplying the end user with whatever format they find useful and/or that will not hinder my sales, even if it means a little extra work on my part. But I get the same results as @BitTree; my photo-jpegs, at twice the bit rate as the H.264, even at a casual glance, look like c**p.

It's no longer about supplying a codec that is easier for the end user to manipulate, it's about supplying them with a clearly inferior file. It just doesn't seem right. You can say it's the fault of Mpeg Streamclip, but I don't think it is. Now my question is, if photo-jpeg should be at least as good as H.264 and I have PrP and AE at my disposal, what am I doing wrong and what should I try next?

Here is the official word from SS:
Quote:

"3. File Format: All clips should be prepared as QuickTime.mov files.

4. Codecs: We prefer footage to be compressed with ProRes 422 / 422 (HQ) or Photojpeg"