I've not read of any quality drop with Titles from CS3 to CS4. Could be there, but just not reported.
Now, how might the Titles in CS4 differ from those in CS3? Are these the same Titles from CS3, and just Imported as a Sequence into CS4? If so, I wonder if the conversion to CS 4 could be an issue?
If these were Titles created in CS4, and inserted into CS3 footage, how do the newly created ones differ from the older ones?
Though for the general creation of Titles, this ARTICLE might give you some tips.
Good luck, and please give us a few more details, to help eliminate variables from your problem.
Thanks for the reply.
I think my talking about importing CS3 projects confused the issue a bit, although that is what I have been doing as I have upgraded to CS4.
If I set up a brand new PAL DV Widescreen 48K project in Premiere Pro CS4 (any version including 4.2) and create some, say, black text on a white background, the letters are soft and blurry. If I do exactly the same in CS3 the letters are sharp and crisp by comparison. If I do the same in CS4 with an AVCHD project then the letters are also sharp and crisp. It is just text in a DV project that is the problem. Please try this. If you don't have CS3 do the comparison between a DV project and an AVCHD project. There is a big difference. You might say that the text is clearer in the High Def project because there are a lot more pixels, but that does not explain why the text is fine, and as sharp as with the AVCHD project, in a CS3 DV project with exactly the same settings.
why the big drop in quality for SD footage titles in CS4 over CS3?
I have not noticed any difference in my own crossover projects.
Do you have your preview set to Draft Quality ?
If you do the fonts will look blurry.
The lower red circle if for viewing your project inside of premiere.
The top red circle is Use Maximum Render for when you are rendering it out.
Im not sure if this is your problem but I hope it is this simple.
Glenn might have pointed out the issue, or at least a potential issue. I had not thought of this.
One additional comment: in some cases, users have gotten better results with Automatic vs Highest, but some have had better display using Highest vs Automatic. While I find this dichotomy odd, these are reports from the field. Highest = best for some, Automatic = best for others? Also, I always use a Display of 100%, rather than Fit, for any critical work.
PS - Jim's comment seems to indicate that he could not find a difference between versions, and he has a very critical eye!