Struggling to make a logo loo sharp…

[Larry Asbell]

The PS file is the exact size of you sequence. It should be larger unless you are using it in the video exactly that size. If you are scaling it or moving it on the screen at all you should be working with a larger version of the logo.

View the Photoshop file at 200%, see the jaggies?

For better results, open the Illustrator file in PS and double or triple the size it’s rasterized to. Save it and bring it into FCP. It will automatically scale back to the correct size but look a whole lot better.

[Bret Williams]

FWIW – FCP is usually horrible at downscaling an image. But I took your eps and imported to PS with a 1920 width. Then brought that back into FCP and shrunk it down and placed it along side of the 100% scaled psd you provided. The result? The downscaled version of the larger graphic does looks a tiny bit better. I think in the end it’s just a bit of a codec issue.

Lower right is the new shrunk back version. Upper left is the original psd you provided…

[Bret Williams]

Definitely a codec issue. Change your sequence codec to animation and look how beautiful it looks. Pristine. How old is that animation codec anyway? I thought prores was supposed to be better. I tried changing the sequence codec to HDV and that looked pristine as well. And then the obvious, change the sequence back to XDCam EX VBR. The logo looked pristine. All standard codecs EXCEPT any flavor of ProRes looked perfect. Go figure.

So there you go. You should work completely native XDCam imo for this project. Here’s what it’ll look like…

[Max Fancher]

Thanks a lot Bret for spending the time on this!! Yah, i’m getting trippy results. Changing the sequence codec to animation makes it pristine. Changing back to XDCAM EX 720p24 looks horrible in the canvas but the exported .mov looks great.

I tried a different way of creating the .psd this time too and that may be helping. Before, I was creating a new .ai from the .eps and then exporting to .psd. This time I opened up a new .psd and made it 2560×1440 (200% – thanks larry!) and then “placed” the .eps into the .psd. Using this new .psd and sticking with the XDCam codec is the best it’s looked yet.

thanks for all your help everyone!!

[Larry Asbell]

I’m real pleased my advice helped you get a good result. I think this shows the importance of looking at how graphics have been prepped as part of troubleshooting how they appear in FCP.

For everyone’s benefit let me clarify my specific point.

Look at the original .eps file. The pixel dimensions are 274 pixels x 127 pixels! That’s the story right there. That’s tiny! The saving grace is that as an eps file it’s still vector-based. That means it’s capable of having beautifully smooth edges at any scale. But you’ve got to change the scale before it’s rasterized, i.e. turned into a pixel-based image.

As we can see from your Photoshop file, 274 pixels x 127 pixels size is about half the size that you scaled it to. Because it looks jaggy we know that you scaled it up in Photoshop where it is pixels. You should always scale it while it’s still vector-based. If you don’t yet know the exact final size then do double or triple what you think you will need, so when you scale it, you’re scaling it down.

When you open an eps file into Photoshop you get a dialog box where you control the rasterizing dimensions. Change the units to read as pixels then enter 2000 into the width. That’s roughly half the width of your 1280×720 frame, multiplied by 3. (Constrain Proportions should be checked, and while your there change the Mode to RGB.)

When I first posted, I didn’t see how small the original size was so I suggested 200%. Actually it could have been 600% but evidently 200% helped enough to make the results acceptable.

[Max Fancher]

Thanks again Larry! My original process was slightly different than you describe here so let me clarify… What I originally did was take the tiny vector-based graphic from the .eps and pasted into a .ai file that was 1280×720 and scaled it up to the exact size I wanted in my FCP sequence. This was still vector-based (i assume) so it should have scaled correctly. Then, I exported it out of illustrator as a .psd so it was still at the 1280×720 size… in FCP it was scaled at 100% and I didn’t adjust that at all.

My new process is to take the .eps and scale the graphic up there and then “place” it into the 2560×1440 .psd which is obviously 200% so when I drop it into FCP it scales it to 50%. Now, perhaps another issue is that the only reason I’m not just dropping a .ai into fcp is that i couldn’t get the transparent background even though it looks transparent in illustrator. I wonder if a .ai would look better than a .psd if I could figure that out?

thanks again!

m

[Larry Asbell]

Max –

Cool. Obviously you didn’t need the rasterizing explanation. Actually your way is purer. I generally grab for PS before Illustrator out of habit.

So, is the lesson that making a graphic double size to scaling it down 50% in FC can look better that at 1:1?

Before we’re sure, let me ask, were you comparing both the before and after as final exports or just in the canvas?

If you compared exports, then I guess making it oversize helped.

[Bret Williams]

Larry- if you look at both my posts you’ll see that scaling the graphic down was pretty much the exact same result as rasterizing it to the size you’ll use. And only in prores where the results were, well, crappy. When I used the native xdcam sequence you’ll see scaling down was not as clear as using the smaller psd at 100 percent. As is almost always the case with FCP. It has a horrible scaling algorithm.

If you took the psd he provided, and opened it in the viewer at 100 percent it wa perfect. But once prores got ahold of it, yuck. Pretty strange.