I discovered a weired behaviour in Resolve on QT export for the web.
If you choose the MPEG4 Video Codec and put the video/data level on Auto, the results are as expected.
in fact Auto chooses video level.
If you forced the MPEG4 to data level, the result looks more contrasty and saturated than the original and there is a lost in DR. However the look remains viewable.
With MPEG4 the setting that preserves the original is Video level (or auto)
But if you choose the H.264 Codec with video/data level on Auto, the result is a gamma shift and the footage looks washed-out and non viewable.
The cure with H.264 is simply to force video/data level to Data(4-1019) and so the footage will be back to normal.
Exactly the contrary of the MPEG4 codec.
With H264 the setting that preserves the original is Data level
As you see, this is the big mess.
Those kinds of details are irritating. For sure they might be a technical explaination an engineer would write a brilliant essay on it.
Fred, I don't see that much difference between your 4 clips. Obviously it's there but a client won't notice.
I also set on export under the left hand pane, the video level to full. That seems to give me the best results.
But, you nailed it. Spaghettification. That’s how this stuff works. They write on top of existing systems, rarely do the makers do full tests on different machines, don’t recommend turnkey solutions like like graphics cards, operating systems and with the 12 million tutorials out there, they all make it seem easy. That’s why I buy my computers from a company that specialize in video.
Nothing is easy in professional digital cinema, even if the screen is tiny.
I needed a new laptop for traveling. I searched, read specs etc. etc. Finally found a new company that specializes in configuring video machines and tests every software, every way possible. I was going to buy a new 2017 macbookpro and they talked me into a custom configured 2015. It has less vram than the newest, but has the fastest ssd made.
We had an hour and a half conversation, they explained why the 2015 was better and even offered to take it back at any time if I wanted to return it for a newer one. So far it’s been amazing and will run resolve faster than the newest, because I duped the data drives, put 100 clips in resolve and in a friends 2017 machine the 2015 went through 1dxII files with motion jpeg codec (a very heavy codec) and ran through 4k red files so fast it was like working standard def and this is on a laptop running a 27” monitor and a broadcast monitor.
In fact I have a 2 year old I mac with 4 gigs of vram, 32 gigs of ram and all ssd, and the macbook pro runs equal or faster and as of the last two weeks never crashed.
But even using my screen shot, check list system, every computer I have works differently. Don’t know why and I have 5 machines in LA, three in the Dallas studio and three in London.
The only way I know is to test files, grade them deeply and conform them. I personally never output less than prores, or dpxhd, then down convert in wondershare for mp4 or h264 viewing on vimeo and even running 2.2 gamma out of resolve, vimeo usually lightens and flattens the file some.
The biggest issue I have is how clients view the files as all monitors and graphic cards react differently. The only “monitor” that seems to make everything looks great is an ipad or iphone. Go figure.
IMO
BC