Any way to decrease the pixilation during dissolves?

I?ve been away from FCP since version 5, at which time we used HVX200?s 720p, and I don?t recall this being and issue.

I got a Canon 5D 1080p now, which should be better quality, but in any case, this isn?t just on the 5D footage, it has the same pixelization problem a still jpg image.

The sequence settings are automatically done by dropping in the clips, and are:

25fps
field dominance none (changed that from upper/odd on the advice of a user, hoping it would help with one of the many problems with fcp7 many users are reporting, namely ?not found? error during render (especially when rendering whole project). Didn?t help, had to render a few seconds at a time.)
1920x1080, aspect ratio HDTV 1080i (16:9)
pixels square
compressor h.264
quality 100%
motion filtering quality normal.

Exported as quicktime, h.264, current frame-rate, best quality, key frames every 24, 1920x1080 HD.

The screengrabs are mid-dissolve (to/from black)
First is from canvass (where it already looks pretty crappy) and second from the quicktime file (much crappier).

This is just a still image (including the flare)





And one shot from the 1080p video:







We are getting ready to shoot a feature on the 5D, so any advice as to what is causing this problem would be greatly appreciated.
Best,
wv

> The sequence settings are automatically done by dropping in the clips
> compressor h.264

There's your problem right there. H.264 is not an editing codec; it is a super-compressed delivery codec, and using it for your Sequence will compromise everything you put in. Convert your clips (and then your Sequence Settings) to ProRes 422 or ProRes HQ before editing.

I usually tell people that if they're relying on dropping a clip onto a Sequence to set their Sequence Settings, they need to stop doing that. As an editor, you should be aware of your Sequence Settings and know why you're picking them, rather than try to let the computer make bad decisions for you. Avoiding rendering is a bad tradeoff for compromised images.


www.derekmok.com

thank you much.
changing the sequence compressor is easy, but the clips? there are hundreds, how would i go about doing that?
apart from the dissolves, the d5 stuff looks good (not color-corrected yet):



best,
wv

You can try converting the h.264 footage and relinking them. Issue here is that I came across a weird bug where it only reconnects 1 clip at a time.



www.strypesinpost.com

thank you for posting.
i converted the sequence to prores and it already improved considerably.
i will try to convert some of the 5D clips in compressor to the same and see what further difference it makes.
even if it couldn't batch-reconnect, i could use the method on the next project.
best regards,
wv