[ This tip was first published in the October, 2004, issue of
Larry’s Monthly Final Cut Studio Newsletter. Click here to subscribe.
Updated May 2005. ]
Preparing still images for video probably deserves a book; maybe two books. It is surprisingly complicated.
We’ve already covered, during the last two months, issues of square vs. non-square pixels. So, this month, I’ll tackle other issues that make creating images for video different than creating images for computer display.
The most important point to keep in mind about transferring still images to video is that all video, all the time, all around the world, forever and ever is 72 dpi.
Video is a low-resolution medium, especially when compared to print.
Consequently, all those beautiful images that you scanned at 300, 400, even 600 dpi are gonna look pretty grim when you bring them into Final Cut — if they can even be imported at all.
This is because Final Cut only looks at the total number of pixels across by the total number of pixels down. The maximum size image that can be imported into Final Cut is 4,000 x 4,000 pixels. However, I’ve seen Final Cut have problems with image sequences where the image size exceeded 2,000 x 2,000.
So, the first thing you need to do to prep your images is open them in a graphics application, like PhotoShop Elements, and check the overall image size — in pixels. Resize the image so that it is no larger than 2,000 x 2,000 pixels; preferably much, much smaller!
Second, the image size of DV video is 720 x 480. This means that a full screen image doesn’t need to be any bigger than 720 x 480. (The image size of SD video is 720 x 486.) However, just to keep us humble, video uses non-square pixels and computers use square pixels.
So, to properly scale images so that they import at the correct aspect ratios, you need to scan your images so that they are 720 x 540 x 72 for DV projects and 720 x 547 x 72 for SD projects.
These numbers used to be 720 x 534 for DV. What happened is that in FCP HD, Apple changed how it calculates the aspect ratio for imported images, resulting in a change in image size for scanned images.
By the way, these pixel values only apply to single-layer images. Multi-layer PhotoShop documents should be sized at 720 x 480 x 72 (720 x 486 x 72 for SD). Sigh…
On most computer monitors, the gamma setting (which is geek for “mid-tone gray-scale value”) does not match the gamma setting for video. This means that images that look perfect on the computer look washed out on video. And video images look exceedingly dark on the computer.
The easiest way to fix this is to create great looking images on your computer, then, as a last step, open the Levels adjustment in PhotoShop or PhotoShop Elements and set the gray setting to 1.2. This darkens your image to match the standard video setting.
Your computer can create colors that a TV can not display. This is due to the computer using RGB colors, while video uses YUV.
The easiest way to see if your colors are TV safe is to import the graphic into Final Cut, then look at it on the Vectorscope (Tools > Scopes) then choose Vectorscope from the pop-up menu.
Colors which extend outside a rough rectangle drawn by connecting the six color reference boxes are colors which are not TV safe. Go back into your graphics application and tone down the saturation levels until the colors are safe.
Sharpening is the process of enhancing the edges of an image to improve the illusion of focus. Some sharpening is OK. However, too much sharpening will cause your images to start to “vibrate” on the TV screen. There’s no hard and fast rule here. However, the best idea is to go gently on the sharpening until you can look at your images on a video monitor to make sure they look OK.
Moiré is the false colors that seem to appear around patterns of closely spaced stripes or dots. The easiest way to prevent moiré is to scan your images at the correct size. Moiré most often appears when you take a large image and scale it down in Final Cut. It’s the scaling that creates the moiré pattern.
If you can’t resize the image before importing it into Final Cut, you can improve it’s apparent clarity by blurring it slightly. I know, it seems counter-intuitive. But, blurring softens the edges of the stripes or dots so that they don’t create a moiré pattern.
The best way to do this is to apply a Gaussian Blur filter inside Final Cut and set it to a value of 0.5. Your moirés should disappear.
If you haven’t guessed by now, I’ll make this as plain as I can. The only way to judge the quality of an image is on a video monitor. Not a computer monitor and not a TV set.
TV sets have all kinds of inherent problems and vary wildly by age and manufacturer.
Computer monitors are designed to properly display computer graphics, not television graphics.
Only a video monitor can show you the effects of interlacing, YUV color space, lower gamma settings and decreased resolution in a way that you can decide if your graphics look good on TV or not.
If you are serious about the look of your projects, you need to invest in a video monitor. However, there’s no reason to spend extra bucks for a big monitor. The best monitors are around 14″ in size. Why? Because whether you view your images on a large or small monitor, all TV images are always 720 x 480 (or 486) x 72. So, bigger monitors just make the pixels bigger.
For me, I have a 14″ monitor in my project studio that works just great and shows me every little thing that’s wrong with my image.
Which is exactly why I bought it.
Here is a summary of what you need to know to prep your graphics for video :
- All TV graphics are a fixed resolution, generally described as 72 dpi
- Scan your images at 720 x 540 x 72 for everything you want to show full-screen
- TV gray-scale values (gamma settings) are darker than on a computer. So darken your computer images by setting the gamma setting to 1.2
- There are colors you can create on your computer that you can’t show on TV. These include saturated yellows and blues. So use your vector-scope to stay safe.
- Don’t over-sharpen images.
- Images that have a moiré can be improved by adding a small (0.5) Gaussian blur filter in Final Cut
- The only way to judge an image is to look at it on the video monitor, not a computer monitor or TV set.
Ben Balser writes:
Two questions I’m trying to clarify. I’ll probably sound like an idiot, but there are some FCP issues I see discussed a lot that I don’t have. And I’d like to understand what is happening with these issues for other users.
One thing I’ve not seen mentioned on the whole PhotoShop issue that I wanted to share. Personally, I’ve never had problems with PS files in FCP. But I am aware others have.
According to the Apple Pro Training Series book on FCP-HD, the advanced one, if you do text in PhotoShop, ESPECIALLY if you do effects like drop shadow, etc, you have to rasterize the text layers. It has to do with the unique way PhotoShop handles text layers.
So this is Apple’s official word on the subject. I just rasterize my text layers (NOT the same as flattening all the layers, just the effects with text), Save As with a “-r” at the end of the name of the original, import that into FCP, no problems.
Let me qualify this by saying I rarely get PS files from a PC, and when I do, I ALWAYS open in the Mac’s PS and Save As according to Apple’s specs FIRST. I haven’t had problems. Bearing in mind all the text layers from PS-CS have been rasterized, as Apple suggests. Also, let me add that 90% of the text in our projects come from FCP’s 3D text generator, or much more often from LiveType. It is rare I get PS files for text. So I may be missing something others work with much more often.
I also get both TIFF and JPEG files from graphic artists to use in our video projects. These are almost always very high resolution in both TIFF and JPEG and I’ve never had a problem with any of them. I’ve done test by down-rez’ing some. Here’s what I am trying to find out. Do high rez TIFF or JPEG make a difference? Does FCP-HD, as some swear by, convert them to 72dpi (thus lowering resolution anyway)?
What I’ve found is, at very high rez, FCP-HD on our systems has no problems. In fact, if I go into PhotoShop and down rez them to 72dpi first, they actually do not look as clear in our final output. Also, I found no difference between TIFF and JPEG. I know there’s a technical difference, but on our final products, both are just as clear. Could that be due to the 800 plus dpi files I’m getting?
We have a jewlery client who has professional promo photos done. We have to zoom into these to show detail in the diamonds and settings on ear rings, necklaces, rings, etc. throughout the videos we produce. Maybe less than 10 or 15 at the extreame most. A 72 dpi photo will NOT give us the detail we need for that close of a zoom, period. Not an oppinion, what I do daily. A 300-600 dpi photo WILL give us extreame detail doing these same 200+% zooms. That is my only point.
So my question is, why recommend lower rez TIFFs for FCP? I’m sure there’s a reason, but I haven’t found it in my tests. And I’d like to not mislead anyone with advice on this issue. There is obviously something I’m missing.
Keep up the great work. Love the newsletter!
Larry replies: Ben, thanks for all your emails. I’ve only summarized them here.
Your key issue is zooming. And you are correct. If you plan to zoom into a picture, you need to bring it in at a larger size than if you simply want to display the image full-screen with no movement.
All video, always and everywhere, is fixed resolution. We most commonly call it 72 dpi. So, none of your high-res pictures are actually high res in video. They are ALWAYS down-sampled – however, that down-sampling occurs during rendering, not import.
Because all video is bitmapped and at a fixed-resolution, I’ve found it easier to tell my students to look at the total number of pixels across and down, rather than the dpi.
Also, if you want to zoom in, you need to import an image at a larger size than full-screen so that you have room to zoom without pixelization. For me, if the image is full-screen with no moves, I import it at 720 x 540 x 72. If I plan to pan or zoom, I import it at 1440 x 1080 x 72. This allows me to zoom to double the image size without problems. Image quality is good.
FCP has an internal resolution of 4,000 pixels across by 4,000 pixels down. So a truly large picture can’t even be imported. I’ve seen many problems with pictures greater than 2,000 x 2,000; especially when you are building a lengthy photo-montage.
Update #2: Tom Wolsky reminds me that the 4,000 pixel limit was removed in version 4. However, using very large images is VERY RAM intensive, so be sure to adjust your memory allocation in System Settings to handle very large images.
Next, most JPEGs are heavily compressed causing compression artifacts. To prevent this, I’ve just adopted my workflow to always use TIFFs. Hi-rez JPEGs should work OK, because the amount of compression is minimized. However, for me personally, I prefer TIFFs.
Finally, at the NAB roll-out of FCP5, Brian Meaney made a big point of mentioning that scaling has been significantly improved in the new version. This may be a good reason to consider upgrading, if true.
2 Responses to Prepping Still Images for Video
I need information regarding scanning images for 4K video projects. Is 72 dpi still applicable? Also, what size should I aim for when requesting a digital photo?
First and most important, IGNORE DPI when creating images for digital video or the web. The only thing that matters is total pixels horizontally and vertically. So, I tell people to set this to 72 dpi, then ignore it. Or 300, or 4,241… It doesn’t matter.
Second, always ask for the largest digital photo they can provide. Bigger is better. Images can always be scaled in Photoshop – or the image editing software of your choice. The worst situation is to get an image that’s too small.
Third, for UHD, you want images that are at least 3840 x 2160 pixels. For “True 4K” you want images that are 4096 x 2160 pixels.
Hope that helps.