|
|
photos to HD
June 14, 2006 06:17PM
|
Registered: 17 years ago
Posts: 1
|
I have this guy who has 4000dpi photos he wants to edit to HD. Not a problem. We'll just drop the dpi down. But to what? My manual (FCP Manual) says all video is 72dpi yet this guy at a local video gear company told me 300 dpi is what I should use for HD. So what do you guys think. 72 dpi for photos or 300 dpi. I thought since video is 72dpi all that matters is width and height dimensions being large enough so you can scale in and reposition...
gomez
Agree with Kevin on this one. What's going to matter more than your dpi, is your w x h. As long as your width and height are sufficient (sorry, not sure what that needs to be for HD), then 72 dpi will work fine.
Now to start up the debate on whether to use jpeg or tiff 
Steven
Sometimes the obvious is hidden in plain view.
Hi,
It always seems to be the never ending story about dpi. dpi is not relevant for video - except you may want to compare the quality of your monitor.
As mentioned above the real size of a frame is relevant. So for HD you will need 1920x1080 (square) pixels for a still. If you want to zoom in for 200% you need twice the resolution and so on.
The only thing where the dpi also may come in, is a bad behaviour of some video applications, which really take this into account, if the dpi is not set to 72.
Regards
Andreas
Some workflow tools for FCP [ www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[ www.spherico.com]
Guy, you need to update your info. We use rich scans in FCP, Movingpicture and AE all the time, to support zoom ins which stay sharp.
I developed a simple system based upon SCANNING DPI- which translates to VIDEO PPI. For our purposes, they are equivalent and it;'s a convenient way to prepare any width of flat art for moco, not simply an 8 X 10 scanned at 72 DPI.
No one however argues that tugging aorund large scans in FCP can be dangerous and taxing to the CPU. These days, because of HD, it's best to bring each in, animate it, export the animation as a sequence-native video clip, reimport that, store the original animated scan file from your timeline in a "factory clips" folder and replace it with the native re-import.
Other than that, the system works for all kinds of flat art widths and you know exactly what zoom enlargement you're scanning for, in any video format.
You don't have to use it. But hundreds do, and I don't want anybody left out. ;-(
www.lafcpug.org/tutorials/basic_scanpro.html
As for moco tools, there's plenty of discussion on best practices for animation. I like MovingPicture plug-in.
- Loren
Photo scan rates demystified!
ScanGuide? Pro compact reference now at
the LAFCPUG Store!
Loren Miller wrote:
> Guy, you need to update your info ?
Did you mean me?
Andreas
Sorry, I'm not trying to insult anyone, Andreas! Only shock people and help.
I mean anybody using an inefficient method for calculating photo scanning for moco. And specirfically, the erroneous statement that DPI has no meaning in video. I think my biggest convert to a "different" way of thinking about scanning DPI is Phil Hodgetts, a guy I respect quite a bit. The method you and Kevin and others describe-- reaching a specific frame dimension, is an unfriendly way to approach flat art scanning for video.
Specifically:
[So for HD you will need 1920x1080 (square) pixels for a still. If you want to zoom in for 200% you need twice the resolution and so on.]
It's not that this is really incorrect, I argue it's the wrong way to approach the problem. Knowing the final frame size of my sequence helps very little in determining how rich to scan a postage stamp. You can get there, but there's a more convenient, 21rst century way!
I learned and developed the ScanGuide? the hard way. In the real world, we have wallet photos, postage stamps, and tabloid newspaper front pages-- whatever producers and directors throw at us-- which we want to bring into SD or HD sequences. We want to keep zooms sharp, which requires rich scanning-- no way around that-- yet avoid needless rendering time.
People who don't do a lot of this get the idea that everything must be scanned at 72 DPI. Then, following this logic, forgetting the ultimate requirement of the video dimension as you correctly indicate, they try, for instance doubling it for a 200% zoom in.
Try that on a postage stamp. You will certainly achieve a 1920 X 1080 frame. It'll be very crisp. With a lot of black around the stamp when you import it. Or, depending on the import "rules" you'll get a stretched scale-up along the longest edge, or if scaled proportionately, just a bad-looking image. You see?
The missing step is *how* to fill a 1920 X 1080 frame with a postage stamp.
You need to scan a specific art width deep enough to fill the video frame. Once you arrive there, then YES, you need to double that to support sharp 2X, 3X, etc enlargement. Then we'r eback on track and it's easy to figure.
But one shouldn't apply a general rule to scanning for video in the real world-- it should relate *what* you're scanning. That's my real contribution. If you're interested, it's a simple system I want you and anyone else to try. You'll be ready for anything.
Go here for the free "Junior" table:
[ keyguides-1.home.mindspring.com]
Scroll down for the downlink link. For the sake of argument, try it. Tell me what you think.
[The only thing where the dpi also may come in, is a bad behaviour of some video applications, which really take this into account, if the dpi is not set to 72.]
I think this is the point made in my post above. FCP, Avid, AE, MovingPicture and other tools not only allow rciher-than-72-DPI scans, they *need it* to support sharp enlargement. But only up to a point; it varies from machine to machine. Until recently, FCP carried QuickTime's 4K pixel limit. Avid supported 5K. MovingPicture supports 8K square. After Effects is the king, well beyond-- what is it now? 44K square? probably higher to support HD. Scan rich, or otherwise, the rendering engine will invent data from nearest pixels, which softens the image as you move in.
So the other point to make about this approach: it gives you only what you need and no unnesscessary data to haul around or render in your NLE.
Some engines are better than others, and you can get away with scanning even less for HD, sometimes by 25% of the ideal scan-- there is so much data! I saw this in my HD tests at WBGH five years ago. (The postage stamp was Elvis!)
SD is less forgiving, based upon my tests, and rich scanning is much more critical.
Enjoy. Everybody stay cool, it's terrible out there.
- Loren
Today's FCP 5 keytip:
Preview effects sections with Option-P or Option-Backslash!
The FCP 5 KeyGuide?: a professional placemat.
Now available at KeyGuide Central:
www.neotrondesign.com
The method often advocated in here is to scan at a high DPI to allow for print work, posters, web etc., then create 72dpi versions for editing. Since the guy in question already has scanned versions at 4000dpi, I don't think scanning is much of an issue anymore. As long as he saves his 4000dpi originals.
By the way, I meant to address "Guys" above, not Guy, giving Andreas the feeling I was singling this person out! Sorry again. My typo.
Derek gives an example I can reverse engineer:
4000K resulting from a scan of what size artwork? It does matter.
Let's say it's an the ideal 8 X 10. (To fill the HD frame edge-to-edge the image area will actually be 5.625" X 10", accurately indicated in ScanGuide? Pro-- not the freebie.)
Going where?
Let's say into an HDCAM sequence, 1920 X 1080.
To do what?
Let's say, producer wants a zoom to the eyes, about 5X required.
You have a 4000-pixel wide scan. Other than the fact that some systems will choke at 4K (recent past FCP QuickTime) and begin to give the bad behavior Andreas speaks of, you have enough to enlarge that 8 X 10 no more than 2X before evidence of softening.
Now what? Use it anyway? It might work out.
Or you can scan intelligently, approach the task from the artwork size and adequate DPI, not the delivery frame size and 72 PPI.
Let's say I don't have an ideal 8 X 10 here. I have a 5 X 7 landscape print. In my NLE I want to zoom into the eyes, 5X.
In the table assembled for HDCAM format, I simply find the width- 7" and I see two things immediately-- 1) how much of the image will be masked in the format, and 2) the baseline per-inch scanrate required to fill the format's frame sharply, no guesswork or blanket DPI scans.
In this case each inch of the artwork is scanned at 275 DPI. Depending upon the qualityof the codec, you may be able to trim this rate, and reduce your render time, but by the book, that's an ideal rate.
Now my finger reads across to the 7" width 5X column because that's the enlargement my client requires. There is a number there, 1375. That is an ideal DPI scanrate to support everything from full frame to 500% enlargement-- sharply. No data invented, and experimenting, you may be able to trim data from it (downsampling in Photoshop to lighten the load).
So my Photoshop image size will be 1375 X 7"= 9625 pixels wide. I'm estimating that'll run about 150 MB as an RGB TIFF. It's big!
You will *not* be importing this directly into FCP. Nor anything else but After Effects! For now.
How rich you must scan helps determine your best workflow.
I cannot install Motion 2 on my machine. Does anyone know if Motion 2 carries a greater graphic import limit, and if so, where is it covered?
- Loren
Today's FCP 5 keytip:
Preview effects sections with Option-P or Option-Backslash!
The FCP 5 KeyGuide?: a professional placemat.
Now available at KeyGuide Central:
www.neotrondesign.com
Loren Miller wrote:
> Sorry, I'm not trying to insult anyone, Andreas! Only shock
> people and help.
I didn't feel offended. (see PM)
> I mean anybody using an inefficient method for calculating
> photo scanning for moco. And specirfically, the erroneous
> statement that DPI has no meaning in video.
I still think dpi does not have any meaning in video. If it would have you would be forced to take into account every output media (monitor, beamer) as with print.
> I think my biggest
> convert to a "different" way of thinking about scanning DPI is
> Phil Hodgetts, a guy I respect quite a bit. The method you and
> Kevin and others describe-- reaching a specific frame
> dimension, is an unfriendly way to approach flat art scanning
> for video.
Not really if you accept that video is dpi-independent. See below.
> Specifically:
>
> [So for HD you will need 1920x1080 (square) pixels for a still.
> If you want to zoom in for 200% you need twice the resolution
> and so on.]
>
> It's not that this is really incorrect, I argue it's the wrong
> way to approach the problem. Knowing the final frame size of my
> sequence helps very little in determining how rich to scan a
> postage stamp. You can get there, but there's a more
> convenient, 21rst century way!
Finally my statement is correct, but you're right: some people may have problems to bring the world of video and flat art scanning together - independent from the century ;-)
> I learned ? yet avoid needless rendering time.
>
> People who don't do a lot of this get the idea that everything
> must be scanned at 72 DPI. Then, following this logic,
> forgetting the ultimate requirement of the video dimension as
> you correctly indicate, they try, for instance doubling it
> for a 200% zoom in.
That's why I said video has nothing to do with dpi - but with final resolution in pixels.
> Try that on a postage stamp. You will certainly achieve a 1920
> X 1080 frame. It'll be very crisp. With a lot of black around
> the stamp when you import it. Or, depending on the import
> "rules" you'll get a stretched scale-up along the longest edge,
> or if scaled proportionately, just a bad-looking image. You
> see?
Not really. If I scan a postage stamp I will achieve a 1920 X 1080 frame by setting the correct scan resolution (and crop or get black borders at some state of zooming). My scan software will tell me.
> The missing step is *how* to fill a 1920 X 1080 frame with a
> postage stamp.
I don't know about US postage stamps that much, but most of them I know will not fill a HD frame because of the ratio - newer german ones though tend to have a "theatrical" ratio which makes things easier (just a joke).
> You need to scan a specific art width deep enough to fill the
> video frame. Once you arrive there, then YES, you need to
> double that to support sharp 2X, 3X, etc enlargement. Then we'r
> eback on track and it's easy to figure.
That's what I said. Go to the needed size in pixels, either 1920 in width or 1080 in height.
> But one shouldn't apply a general rule to scanning for video
> in the real world-- it should relate *what* you're scanning.
I totally agree - the needed (final) image dimensions in pixels is the thing which is constant - dpi will vary depending on the original pictures size when scanning them. But here is where dpi might really come in, but from the side of the picture source. You may want to keep the original dpi raster from the stamp (or news print etc.) or you may not. But that's another theme.
> That's my real contribution. ? and rich scanning is
> much more critical.
> Enjoy. Everybody stay cool, it's terrible out there.
I think the suggestions Kevin and others (including me) made are not that different from that what you said. We all end up with the same result.
All of us want to help people out (and help them to understand). Maybe the others like me had been to short with the answers.
And yes? we stay cool.
Kind Regards
Andreas
Some workflow tools for FCP [ www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[ www.spherico.com]
Andreas-
[Not really. If I scan a postage stamp I will achieve a 1920 X 1080 frame by setting the correct scan resolution (and crop or get black borders at some state of zooming).
And what is the terminology for scan resolution? Dots per inch: DPI. So... how is DPI irrelevant? ;-)
[My scan software will tell me.]
More on that interesting screen shot you sent me in a second. It has potential.
[I don't know about US postage stamps that much, but most of them I know will not fill a HD frame because of the ratio - newer german ones though tend to have a "theatrical" ratio which makes things easier (just a joke).]
We have those too. Let's just assume I have a 1' wide postage stamp. I'm putting it on my scanner bed. I'm opening my scanner software to advanced mode to control whatever I need to control.
I know what video format I'm ultimately scanning for. let's say DVCPRO 1280 X 720. What do I do? What DPI setting to fill the frame? If you say 72 DPI you're dead wrong. To support a sharp zoom in? If you say double or triple that, you're dead wrong again! Right?
Feel free to post any scanning system you like. I don't monopolize this space. Martin Baker of Digital Heaven is another friendly acquaintance and he came up quite independently with an elegant Flash-based DPI scan calculator which follows the same approach as ScanGuide Pro-- it's based on "What's your photo size?" On the other hand, hundreds of folks are happily using my little reference booklet and I'm very proud of that-- it's carries a bit more detail and art range and other useful info.
[dpi will vary depending on the original pictures
size when scanning them. But here is where dpi might really come in, but from the side of the picture source. ... But that's another theme.]
LOL. Varying the DPI is quite *central* to the result you need. It sounds to me like you understand my approach-- "from the picture side," as you say-- solving the problem based upon the varied size of flat art coming across your workdesk, not general rules of thumb of video framing based upon 72 PPI that primarily confuse average users.
[I think the suggestions Kevin and others (including me) made are not that different from that what you said. We all end up with the same result.]
We do, mostly, but getting there is not as comprehensible. I know, I used to be in that camp. Here, a systems approach to scanning for video requires that you deal with art width and relate that directly to dots per inch in your scanning software, not general doubling of SD or HD frame sizes, which don't take into account the size of the original work and need to be varied, an awkward process. A systems approach is far friendlier than general "blanket" scanning settings and needless rendering.
[All of us want to help people out (and help them to understand). Maybe the others like me had been to short with the answers.]
And I tend to go on too long. Thanks for sticking with it!
- Loren
Today's FCP 5 keytip:
Preview effects sections with Option-P or Option-Backslash!
The FCP 5 KeyGuide?: a professional placemat.
Now available at KeyGuide Central:
www.neotrondesign.com
Hi,
Maybe the misunderstanding was that 72 dpi, I was never talking about 72 dpi but I was talking about final image dimensions - what ever dpi they might have.
Regards
Andreas
So below a link to another approach to figure out image dimension and dpi.
[ www.spherico.de]
Some workflow tools for FCP [ www.spherico.com]
TitleExchange -- juggle titles within FCS, FCPX and many other apps.
[ www.spherico.com]
I got lost I need to add scan images from a 6 by 5 negative to my hdv secuence?
anybody can help and straight up tell me how to scan them i got lost with all the technicalities...
I will no zoom in more than 150 and mostly add movement to them
thanks
J.
Ps: will animate in fcp...
Hi,
It's simple.
If you need to zoom 150 percent with HDV, you need an image width of 1920 pixels x 1.5 = 2880 pixels. Having a 6 inch negative means the 2880 pixels/dots have to be distributed to the 6 inch wide original. This results in 2880 dots / 6 inch = 480 (dots per inch). So that's your scan settings.
Regards
Andreas
I'm trying to relate this to FCP movie editing and the use of still requirements. Again, I'm making assumptions so I can learn and get corrected, because I don't know too much about this subject.
If I want to use a still in my movie project (whether SD or HDV), isn't it better to stick with the relevant pixel format? (720x480 and 1920x1080)
If I have to zoom in on the still, then I should bring in the still at the proper multiple of the frame format I'm using. Right?
Coincidentally, I shot a still with my Nikon film camera and then worked on it in Photoshop. The slide was scanned at its normal size, a little over an inch in height... I scanned it at 1200 dpi. When I outputted the picture for printing, I changed it to jpeg cmyk (I figured that was best for printing) -- I wanted to end up with an 8x10 head-shot. I'd never done this before. The resulting file was 80 MB! Kinko's couldn't make a copy from it :-)
I went to photo lab and they said the size of the jpeg was too small. They wanted to charge me for enlarging it. I said I had a computer and could do that myself. They said all I needed was 300 or 400 dpi. So I ended up making a 400 dpi file and I enlarged the photo to 10 inches by 5.5 inches approximately. This yielded a good enlargement, and the file was 1.7 MB.
Did I do it right?
My question here, with respect to using this photo in FCP, is that perhaps I should make a 720x480 reduction in the size of the file and reduce the image size to 72 dpi in Photoshop.
Andreas writes-
[It's simple.
If you need to zoom 150 percent with HDV, you need an image width of 1920 pixels x 1.5 = 2880 pixels. Having a 6 inch negative means the 2880 pixels/dots have to be distributed to the 6 inch wide original. This results in 2880 dots / 6 inch = 480 (dots per inch). So that's your scan settings. ]
Correct, but it's even simpler!
Knowing your delivery format, find your image width in ScanGuide Pro, (6"  look at the baseline scanrate of 320 listed for 100% width display (means it fills the frame width), and multiply that by 1.5. Same result. You don't even need to do that if you want to zoom in sharply to whole number factors like 2X, 3X up to 8X for HD 1080. Scanrates are conveniently listed for these multiples.
This is what I mean! Same result, different approaches, that's all.
I'm lousy at math. ScanGuide is a simple systems approach deriving from the real-world artwork piled up at your desk and ready to scan. A systems approach works for anything you throw at it (within a reasonable range of 1" to 14" wide art-- after that, begin doubling existing widths) and you don't need a calculator, just a desk ruler. Simply turn the page to the delivery format you're working in and find the ideal scanrate for your art width and creative intent.
Again, the DV ScanGuide "Junior" is a freebie on my website, listed above. Grab it! Try it.
Filmman writes-
[My question here, with respect to using this photo in FCP, is that perhaps I should make a 720x480 reduction in the size of the file and reduce the image size to 72 dpi in Photoshop.]
Until I knew you were working with an 8 X 10 I couldn't possibly answer the question. It sounds like a typical portrait mode headshot-- 8" width.
What do you want to do with it? Fill the frame 100% width or frame height? Use 8" width if you don't mind cropping top or bottom, because scanning for 8" width means masking to 6" height-- these reticle dimensions are indicated in inches in ScanGuide Pro, not the Junior guide.
If your creative intent demands full artwork 10" height in the frame, ScanGuide indicates what that 10" height translates to in width-- in this case, about 14". Scan 52 DPI (as if your artwork is 14" wide) and your 10" art height can be scaled to fill the frame height. You will also have black borders along the image left and right. It'll be sharp and it'll be a lighter weight file than if you'd scaled down an 8" scanrate to fit.
What format? In DV-NTSC 720 X 480 you can fill the video frame 100% by scanning the 8" art width at 90 DPI.
In HD 1920 X 1080 you'd scan the same art width 240 DPI to accomplish the same task.
Neither case requires the whopping 1200 DPI you indicate. You might scan 1200 DPI to zoom in 5X on that HD head shot, and you know what? I've seen great results by scanning less, so I always advise experimenting when scanning for HD, it really depends on the codec and rendering engine. But by the book, that scanrate supports 5X enlargement.
[They said all I needed was 300 or 400 dpi. So I ended up making a 400 dpi file and I enlarged the photo to 10 inches by 5.5 inches approximately. This yielded a good enlargement, and the file was 1.7 MB.
Did I do it right? ]
If it looks good on video you did it right. Now, can you remember what you did and why? What about the next photo, a 3 X 5 in landscape mode? Or that wallet photo, 2 X 4? if you're a polymath, have a ball.
The other thing I emphasize these days is, once you have your rich scan in the timeline, once you've keyframed and rendered the animation of the .psd, .tff or (ugh) .jpg RBG file, export the clip as a QT movie in the native timeline format, reimport it as a simple movie clip and replace the rich scan, storing the scan file as a "factory clip" in some folder in the Browser or even offline on a CD or flashdrive. Loading your timeline with dozens of "live" keyframed rich scans is an invitation to a Poseidon Adventure.
I imagine Motion can dust off photomotion with ease but others must comment here. If your system supports Motion-- older machines and OS's don't-- use that or some other photo animation program and round trip. Of course, if you're trying to time animation precisely to sound track you may be going through a revolving door, unless you can import the track (or at least markers) for timing.
- Loren
Today's FCP 4 / 5 keytip:
Do a virtual Audio Mixdown to lighten playback load with Command-Option-R!
The FCP KeyGuide?: your power placemat.
Now available at KeyGuide Central.
www.neotrondesign.com
|
|
All the news now and in your digital future
