In all the discussions I've seen little comment on image quality. My initial impressions are that X delivers a vastly improved output for web delivery, sufficient enough I decided to edit my latest project using it. In a test I took a Fcp7 project I had done and delivered to the web, rendered a ProRes master, plunked it into a timeline in X and sent it to YouTube. Now maybe it was my compressor settings in Fcp7, I don't know. But there were far fewer compression artifacts, the image seemed sharper, the gammas were spot on, basically a WYSIWYG experience I've never achieved in the past. Or did I just have terrible compressor settings in Fcp7? Anyone else do any comparisons?
How can you make a quality judgement based on the two default export presets within the NLEs without knowing the real settings? And yes, exporting directly from FCP7 to a compressed web format has always been crap.
Oliver Peters Post Production Services, LLC
What I've done is gone to a variety of computers and watched. Not very scientific I know that's why I'm curious if anyone has done a more scientific analysis.
Good topic! Not just for web delivery but in general.
I am also busy trying out 2 other NLE's and have found some very interesting things regarding image quality. Since our company does titling of programs that we dub in Dutch, I find it very important to do not degrade/transcode the rest of the program which we don't touch.
FCP 6 & 7 have the nice ability to save a quicktime in the same format as the sequence, when a file in that sequence is just a straight cut and matches the sequence settings, saving to a new file (or in my case wrap to MXF with MXF4mac) is just a copy and doesn't degrade the image quality.
I think this is not the case with FCPX because (while you can use ProRes as its native codec) it will handle most codecs in the timeline directly, so the output of a Quicktime file has to be rendered to one specific format. Another NLE which I won't name (yet, because I am not completely done testing), also uses that same philosophy.
At the moment I think the image quality that FCPX outputs is better than the other NLE (probably not better than FCP7 because that leaves the material completely intact, still need to compare). The test I am doing is rendering out the same scene a couple of time, each time putting the new render in a new sequence. It's not something you'd do in reallife but I think it's a good test to exagerate what happens when the image is rendered. In my case the formats of choice are IMX and XDCAM because those are the final delivery requirements that I need to create (we ingest in those formats, or recieve mxf's in that format, do the titling and then wrap the deliverable). In FCPX you see the color information gets worse but the picture stays sharp. In the other NLE after 3 passes the picture is looking terribly pixelated! (the 3rd NLE I mentioned earlier still has to be tested for this)
With ProRes, this is happening much less, there is a difference, but I can easily do 10 passes (in FCPX, haven't gone that far in the other NLE yet). It gets a bit less sharp but the colors stay intact quite well. Also in the other NLE with ProRes the outcome seems less worse. I still need to compare. It could also be a bug with the particular Sony codecs. However like I said this bug doesn't happen in FCPX. Only the colors gets worse each pass.
Now to say that I am comparing codecs. I am not. I am just trying out the same test with another codec as well to see if it's a bug with that particular codec. And I am doing the same process in 2 NLE's and there is a difference when using the same codec. Odd?
The main issue that I am having is that when choosing another NLE than FCP7 we will also need to reconsider our captures. Uncompressed is not an option because with the amount of material that we handle we'd need lots of more storage and we already have quite a lot. ProRes is an option (actually we already use it for most HD material because for editting it's just a better format then XDCAM) but that still means we already have to make a complete render for every deliverable, while we can now output one in 5-10 minutes this will probably double in time. I don't like that at all... we have quite a good workflow, so no need to change yet, but in the future?
[Geert van den Berg] "I think this is not the case with FCPX because (while you can use ProRes as its native codec) it will handle most codecs in the timeline directly, so the output of a Quicktime file has to be rendered to one specific format."
" In my case the formats of choice are IMX and XDCAM because those are the final delivery requirements that I need to create (we ingest in those formats, or recieve mxf's in that format, do the titling and then wrap the deliverable)."
I'm curious if you select XDCam on the export settings of the share menu does it do a copy of XDCam material or does it process it in any way?
Geoff, thanks for pointing me in the right direction!
I was totally neglecting the fact you can also select the codec under share/export. I was always using the last option, which uses a Compressor setting and which always transcodes. The regular export appears to be working just as in the FCP7. Thank god, this made my day!
However what do find strange that 'use project settings' always transcodes to ProRes, while the first clip I put in the timeline is IMX or XDCAM, but selecting the right codec from the dropdown menu does the trick. The image doesn't loose any color information anymore.
This is a big + for FCPX on my list.
Off course I still have an issue though, and that is that I can't do a partial export from within a FCPX project without going to Compressor... Off course there workarounds but I'd like to keep the correct timecode.
But image quality first, the rest will come (hopefully).