flicker/movement in still images

[Phillip Van west]

No. NO. NO!! Post Man is absolutely correct: DPI is IRRELEVANT (and can cause no end of frustration) when working with video. Let me give you an example:

I have a 4 inch by 3 inch photo I want to scan for video use. If I scan it at 72 dots (or pixels) per inch, I wind up with an image that’s only 288 by 216 (4″x72dpi by 3″x72dpi). Too small for my 720 x 480 NTSC sequence – it’s going to look awful if I have to scale it up to full size. So I need to scan it at something like 200 dots (pixels) per inch to get an image that’s 800 by 600. A little too big if no panning and scanning is to be done, but no harm done, either. The point is that it’s the size of the video frame that matters. The same image, at 720×480 or 7200×4800 will look IDENTICAL on a television.

Get it? DPI is important if you’re working with print media because you really ARE working with dots of ink per inch. With video, it’s just a holdover that some people refuse to let go of, like the concept of IRE in a digital environment – absolutely meaningless. I really hope this helps.

pvw

Phil Van West
Terra Nova Productions
Denver, CO
Video Production/Post-Production

G5 DP 2.5GHz / 4.5 GB RAM / OS 10.4.8 / FCP 5.1.2 / QT 7.1.3

[Tom Matthies]

I disagree. DPI DOES matter. If you re-read the orginal post, there was a problem with severe interfield flicker as a result of TOO much detail and the software doing it’s best to try to interpolate the graphic for use in the video world. Even though a graphic at 720×480 will fill the screen the same as a graphic at 7200×4800, when you go into the motion tab you will find that the actual scale of the larger graphic is a much smaller value; much less than 100%. All this extra information is basically thrown out when the graphic is displayed at the normal 720×480 size on the screen. The software is then given the task of re-interpolating each and every pixel of the original graphic and the regenerating a new, suitable version for display. You have undoubtedly noticed this when working with larger graphic files. Things just tend to slow down and the display takes longer to update.
A graphic at 720×480 and 72dpi is not the same as a graphic at 720×480 300dpi. The later is MUCH larger in terms on information that the programs has to then plow through. Many commom digital still cameras like the Nikon D70s is usually use for stills will yield a Jpeg file that is around 3000×2000 pixels at 300 dpi. While you can use this file in FCP, it will seriously slow down your system while you manipulate these images. Plus the added detail will likely result in a bad case of interfield flickering. The newer G5 based systems dont have as much trouble as older systems, but there is still a noticable difference in the speed at which it will handle these larger graphics. When I use any stills from my camera I normally reduce them down to a bit more managable size. If they will be used only full frame with no pan ‘n scan needed I simply take the original 3000×2000 300dpi file into Photoshop and reduce the dpi to 72. This will give me a file size for FCP that is almost exactly the same as the screen size for video. If I need to do mild pan ‘n scan, I will simply reduce the dpi to 144, exactly double the screen size. This gives me plenty of real estate to move the still around the screen. The DPI of a graphic does affect it’s useability in FCP. Drag that still into FCP that is 2000×3000 at 300dpi and go to the motion tab and reset the scale to 100% and see what happens. You will get the idea.
The flicker experienced above can be the result of too much detail. A poster suggested adding a bit of blur in Photoshop. The same can be accomplished in FCP by adding a touch of Gaussian blur right from the filters folder. A small amount goes a long way, but it generally does help. Or add a bit if vertical only blur in Photoshop to help preserve your horizontal detail. There is definately such a thing as too much detail…when it comes to video.
tom

[Phillip Van west]

I couldn’t agree with you more.

I’m certainly not suggesting that a large image is as easy to work with as a small – of course it isn’t. But it isn’t the DPI, per se, that matters – it’s the size of the image we’re trying to work with, and of course we shouldn’t use an image that’s any larger than we really need.

But my point stays the same: “Dots per Inch” is meaningless in video. Sure, DPI may come into the equation when we’re adjusting an image’s size, but the idea that “72 dpi (or whatever) works best in FCP” is meaningless as well. Wouldn’t you agree?

pvw

Phil Van West
Terra Nova Productions
Denver, CO
Video Production/Post-Production

G5 DP 2.5GHz / 4.5 GB RAM / OS 10.4.8 / FCP 5.1.2 / QT 7.1.3

[Neil Ryan]

I was brought up on Avid systems, where setting stills at 72dpi DID matter. And some people follow that out of habit, like I did until recently.

Now, it is NOT relevant on FCP. Please, read this, like I did, then do the simple test between two pictures with the same frame size, but different dpi settings, like I did …

from page 339 of the current Manual:
Video Is Not 72 Dots per Inch
There is a myth in video graphic design: Since some older computer displays used 72 pixels per inch, all video created on a computer must be at this resolution. This is not true or necessary. The dimensions of a video image are dependent only on the number of horizontal and vertical pixels used in the image. Pixel dimensions alone determine the resolution of a video image. You can easily test this yourself by creating two 720 x 480 images in a still graphics program, setting one image to a resolution of 300 dpi (dots per inch) and the other to 72 dpi. Import both images into Final Cut Pro and compare the two. They are absolutely identical. This is because video editing software does not use the dpi setting of a graphic image.
Even though the dpi setting for your graphics is irrelevant for working with video, keep in mind that many people may still adhere to a policy that graphics for video must be 72 dpi. To avoid confusion with other graphic designers, you can just as well leave your video graphics at 72 dpi. Just know that there is nothing special about this setting.”

I hope that increases people’s knowledge base and helps lessen the load on people’s editing routines.
Cheers,
Neil.

[Tom Matthies]

Once the graphic is into FCP it doesn’t make that much difference as long as it is big enough for your needs and yet not too large to actually degrade your finished product with all kinds of artifacts from the excess resolution. Must graphics come from outside FCP. Out in that world, the DPI resolution does ndeed matter. It’s an important consideration when you product your graphics and especially important if someone else is doing them for you.
And also, please remind those print people; no CMYK if you please. 🙂
tom

[Kevin Jones]

A simple fix I have used for stills is to go into the motion tab and rotate the image a very small increment. (point-zero-one) This fixes shakey stills most of the time.
Good Luck!

Kevin Jones

[Tom Matthies]

Hmmm. Now there’s an interesting idea…
Tom

[Bill Lee]

[staircar] “Thanks for replying so quickly. The images where the shimmer is most pronounced are 2473×1677, 2272×1704, and 2549×1848….. pretty big. I tried the motion blur but it didn’t do anything. I’ll try playing around with the motion blur, maybe increasing it to a certain point, and see if that works. If you have any more suggestions, keep ’em coming 🙂

OK, well I’d agree with Tom. A one-pixel high blur will greatly improve things. Not motion blur – use the de-interlace filter.

What you are seeing is your 30 frames per second (25 in PAL) stills being shown on a 60 fields per second (50 in PAL) display device (you are doing your evaluation on an external TV/reference monitor, right?) Every 1/30 of a second (1/25 in PAL), every second line of your image is being displayed, then 1/30 of a second (1/25 in PAL) later the alternate second line in your image gets shown, then 1/30th of a second later, the first set of alternate lines gets shown. This can cause an obvious and ugly flickering effect, depending of the difference between the adjacent scan lines in your stills. Large difference = big flicker. So, anytime you may get a moir