How to Output the Highest Quality Still Images from Apple Final Cut Pro

OK, this gets a bit, well, perhaps more than a bit, into the weeds. If so, just read the Executive Summary. But, I found these results interesting.

Dave McCollum and I got into a long email conversation about how to maintain the highest image quality when adding still images that we don’t intend to do moves on into a Final Cut Pro video project. Should we import still images at the same size as the project or images at double the frame size of the project?

My understanding was that importing still images at 2X the project frame size provided the best image quality. Dave’s testing seemed to indicate that importing at the same frame size as the project was the best option.

Clearly, more research was needed and I had some spare time this week. Here’s what I found when I tested still images in Final Cut Pro.

NOTE: I’m working on similar tests for Adobe Premiere Pro which I’ll publish soon.

EXECUTIVE SUMMARY

As long as you don’t need to create moves on stills, importing stills at the same frame size as the video project yields the highest image quality. There is a clear difference when images are scaled – they look a bit “blurry” – which can be verified using Photoshop.

• This is true for projects using either ProRes 422 or ProRes 4444
• This is true whether images are manually scaled or use Spatial Conform > Fit
• The higher the resolution of the image, the more image variation on export.

If you DO need to create moves on still images, such as pans or zooms, you must scan or create images larger than the project frame size to retain image quality. (Remember, though, that scaling any still image larger than 100% causes it to look increasingly blurry.)

NOTE: I didn’t test scaling any small images larger to fill the frame size of an FCP timeline because scaling small images larger than 100% always makes them look blurry and pixelated

THE TESTS

This was my source image. It was shot on an iPhone at 4K resolution. It has lots of detail with a full range of grayscale values and good color. Using Photoshop I cropped – not scaled – the source image to 3840 x 2160. This was saved as a TIFF. This became my reference image.

From this source image, I then scaled and exported two more images in Photoshop, one at an arbitrary 16×9 size and the other exactly the size of the project I would use for testing.

• 2688 x 1512 pixels
• 1920 x 1080 pixels

I then created the following projects and exported stills from each:

• 1920 x 1080, ProRes 4444 using Spatial Conform > Fit to scale images to fit
• 1920 x 1080, ProRes 4444 manually scaling each image to fit
• 1920 x 1080, ProRes 422 using Spatial Conform > Fit to scale images to fit
• 1920 x 1080, ProRes 422 manually scaling each image to fit

Finally, I took each set of images back into Photoshop and compared them to the original using the Difference blend mode. Difference highlights differences in both color and grayscale. All images were processed in Photoshop at 100%. They are displayed here at a reduced size, but you can see the full size by clicking an image.

Images which were identical showed black. Differences were highlighted as shades of gray. The more differences between images, the brighter the display. I made no changes to any image in Photoshop.

I ran about 30 tests. Here are the highlights.

THE RESULTS

To make it easier to see differences, I selected a detail of the source image. This same section would be used for all tests.

First, I compared the visual look of the original 1920 x 1080 image to one exported without scaling from Final Cut. Whether put into a ProRes 422 or 4444 timeline, the exported image look identical to the source. There are no image artifacts created when exporting an image scaled to the same size as the timeline when measured using Difference.

Next, I compared a scaled image to the original. Notice the blurriness in the scaled image. This blurriness increased as larger images were scaled.

Next, I compared to see whether it made any difference if an image was scaled manually to fit the frame or used Spatial Conform > Fit. The answer is that it didn’t; images that were the same size all had the same amount of artifacts regardless of how they were scaled.

This compares the amount of change introduced based on the size of the timeline image to an unscaled clip. The less an image is scaled, the closer the image quality matches the source image. Two perfectly matched images would be solid black. Here, the brighter artifacts on the right indicates more differences because a larger 2x clip is scaled more than the smaller 1.4x clip on the left.

This compares differences when exporting the 2X image (which shows the most artifacts) from a rendered ProRes 422 vs. ProRes 4444 timeline. When compared using Difference, there are no differences between exporting stills from a 422 or 4444 timeline, but there are differences created by the scaling itself.

SUMMARY

Scaling images introduces blurriness and other artifacts into still images. The more you scale an image, the greater the amount of artifacts. The quality of using Spatial conform is the same as manually scaling images using the Transform menu.

So, for images that don’t move, create/size the image at the same frame size as your project before you import it into Final Cut Pro.

Thank you, Dave, for suggesting I look into this.

---

Bookmark the permalink.