Best way to create titles going from HD to SD

[Rafael Amador]

[christoph Gelfand] ” The text looks pretty good, but I am doing a test both ways to see if there’s a difference.”
That’s the best. If they look OK after downscaling you can avoid a second rendering.
When you go to downsize your ProRess movie, don’t forget to check “Render Motion Effects: BEST”. The downscaling will be done with more accuracy
rafael

http://www.nagavideo.com

[Christoph Gelfand]

Interesting. I have received different opinions on this in three different forums. One post agreed with what you said, but another said:

If you are nesting an XDCAM sequence in an SD sequence, do remove the text from that sequence, as you don’t want FCP to be scaling the text from the motion parameters in the nest along with the rest of the footage. As a general rule, the motion parameters should not be used for scaling text. Instead, readjust the font size from the controls of the titling tool you are using.

But I do think they look ok…so maybe I’ll save myself the headache. But the principals he mention make sense even if it’s only a marginal difference…

[Chris Poisson]

Christoph,

Sometimes I do it this way:

Set up and edit in an SD project set to ProRes. Edit in your EX1 footage and let FCP scale it. Put in your text and graphics as usual. When all done, change the video processing settings to “high precision YUV” and set motion filtering quality to “best” and under render control set quality to “best.” You will now have to re-render, but all this saves a step and the results are outstanding.

Other times I work in HD and let the Kona card downconvert, but the results look the same.

Have a wonderful day.

[Chris Poisson]

Christoph,

Regarding your original query, the best titles come from vector based tools, such as Boris Title 3d. Plus all the other stuff I mentioned.

Have a wonderful day.

[Christoph Gelfand]

Thanks for the response. That method works, but then you have to render your footage to SD before you edit. Additionally, I can’t seem to make the titles look even remotely good in SD- nor have I ever been truly satisfied with clarity of text in Sd using FCP. For years my text has always looked not as crisp as I’d like it to…

Here’s a screen grab from the downconverted image (from HD to SD)- I’ve blurred the subject’s face for confidentiality.

[Rafael Amador]

Hi Christoph,
[christoph Gelfand] “That method works, but then you have to render your footage to SD before you edit”
No. As Chris wrote, you drop your HD footage in an ProRes SD Anamorphic sequence. Put your graphics and evrything and when all s ready set “Render all YUV material in High Precision YUV” and Motion Rendering BEST.
You have RT performance when editing your HD in a SD sequence. No need to render until the end.
i use the same method for working in EX-1 and ending in SD. The best setting for rendering works great for downscaling HD footage and really fast comparing with Compressor.
As Chris points too the text generator are Vector based applications. In the moment you render they get converted in a ‘Bit Map” format. A vectorial graphic can be resized without any lose of quality, a Bit Map don’t. If you can avoid any resizing, the best. (but this counts for the picture too).
And as Chris points too, use only the Boris 3D.
rafael

http://www.nagavideo.com

[Chris Poisson]

Christoph,

At some point, you’re gonna have to render everything to SD, so in my mind it makes no difference. I have really happy clients using this method.

Have a wonderful day.

[Christoph Gelfand]

Got it. So that picture I posted was created using Boris 3D and exported using the meothd you detailed. Do you think the text looks ok? To me it looks kind of funky…

I’m just trying nail down this text thing once and for all! I find that LiveType files look fine when juxtaposed, but I use Boris for everything not involving motion based text.

[Chris Poisson]

Well,

Really hard to judge on a computer monitor, you need to see the downconversion by WHATEVER method on a good NTSC monitor.

Have a wonderful day.

[Michael Gissing]

Two thoughts. Firstly don’t nest. Copy & paste the elements to the new sequence. The comments about text scaling in a nest may be an issue. Frankly I avoid nesting and there is no need for your application to nest. A non nested sequence should handle the text properly.

Secondly if you were viewing text on a progressive HD sequence (you didn’t say what the XDCam was), then the SD sequence, which is interlaced, will show the text badly on your computer screen. You can only judge interlaced text on an external monitor. Even if this isn’t the case, after looking at HD for weeks, SD looks shabby. Try generating a new text element and compare it to the ones from your original sequence. I bet they look the same. If not, then we would like to know.

I have been disappointed with FCP downscaling so now, I use my HDW 1800 HDCam to do it in realtime. Yes the graphics take a slight hit, but overall the pictures are better. I believe the Kona3 realtime downconvert is also excellent and of course a Terranex would be better.