I'd go for a 4000 dpi or higher (optical resolution) scanner.
That way you've got some margin for cleanup before downsampling to 1920x1080 (downsample in Photoshop - Premiere can slow down trying to deal with larger images).
The only choice in file format is whether you want lossy (JPEG) or lossless (TIFF, PSD, etc.)
Since you should be scanning at 16 bit/channel, that would rule out JPEG anyway.
How are you? Hope the computer is running well!
For 1080 video, the frame size is 1920x1080 pixels, and 720p is 1280x720 pixels.
You scanning app will likely give quality options as DPI, or Dots Per Inch.
So if your slide is one inch square and you scan at 1200dpi, the resulting image should be 1200x1200 pixels.
If scanning an 8x10" photo, then you could use 200dpi and have plenty of pixels for an HD image.
I've only scanned photos for video, and have not done small slides or negatives, so don't have any detailed insight in those areas, other than to say you will need a good scanner with a high DPI precision.
What format? Probably everyone will have their own opinions. To keep quality up, you might want the scan to be uncompressed, such as .bmp or .png rather than .jpeg. Premiere will accept most still image formats. Something to consider - don't make the stills too large, as this can bog down Premiere. For instance, don't import 5000x5000 stills if you only need 1280x720. They can be scaled to exactly the video size needed, or close to it anyway. Of course, if you wish to add motion with pan and zoom like Ken Burns does, then you will want the still to be larger than the video frame. how much depends of how far in you will zoom.
Safe Harbor Computers
I used to scan stuff for people using a Minolta Dimage Scanner.. Nikon has some coolscan scanners for 35mm only that work well..
Here's the options of the dimage scanner...basically...
If a 35mm slide ( mounted or not ) was scanned according to above specs you end up with this...
You would, after doing your psd work, go to 72 dpi for video...making the image basically 3x as large.... in which case you would then downscale it for your video dimensions...
It's been my experience that after downscaling it you need to revisit " levels " to fine tune your downscaled image...
This should help you even with other scanners out there...
I would like to add that once an image has been scanned, forget about DPI. It has nothing to do with video.
We are looking at PIXELS at that point. DPI would only apply to printing the image, which is why in the "Image Size" screen grab shared previously, "Resolution" is shown in the "Document Size" area, also shown in inches. The "72dpi for video" thing is very old-school and was perhaps a rule of thumb for the very small displays that were around at the time. But in any event, forget DPI with video production, it has no place in post, just look at pixels.
Thanks for pointing that out about dpi vs ppi. as it is in fact as posted... a cross between print ( offset press etc ) and new 'computer' type stuff.. but it's so complicated and historical I didn't wanna get into it. I don't think poster needs to get into it really for his / her question.
Some scanners for film will have dpi info though.. so anyway, hopefully this doesn't become an issue to just get the question answered ? I don't know what else to do so I'll just hope the scanner and the output the poster wants to understand come out of this..
duh..I'm an idiot !
Thanks for all of the responses! As I go deeper into the subject I want to add that what I am doing is creating a library which includes scans of all of my slides as well as digital copies of my DV footage. So the file format portion of the question involves how I should package the scanned images for storeage. I presume that I should use a lossless format since I many will be adjusted by Photoshop and used in videos. However, some will probably be printed. I've got some 3000 slides to scan and while .tif has been around for some time I was hoping to be able to get some compression to save on file space. I don't know anything about GPN and wonder what will be the best choice for this archive. Also, I'm not sure I understand the DPI vs PIXELS concept. I'm used to working with prints and need to up my understanding of video resolution.
With a dual purpose, the archiving and then use in a Video Project, I would scan at the highest optical (not software interpolated) resolution, that your scanner allows. The software used, might dictate the format used to Save, but if you scan through Photoshop (Import), then I would choose PSD. If storage space is an issue, then you might consider TIFF with LZE compression, but PSD would be my first choice. If you scan with other software, then a TIFF would be my number two.
Once that is done, you can create a simple Action in PS (or use Image Processing Script), to Scale the Images down to near what you will be using, and use Automate>Batch to Scale and output an entire folder to another folder, the one that you will use in your PrPro Project. I still like PSD (Photoshop's native format, and easily Imported into PrPro).
Just how I do things, and good luck,
Thanks for the response Jeff. I'm looking forward to seeing you at your place this month and seeing your new editing setup. As to the computer, I'm plugging along my learning curve. I've really got too much training info! One thing about the computer which still bothers me is that there is this almost constant sound coming from the big box. I guess I should take the side off and visually inspect. It seems like the sound is produced when the hard drive is doing something. Haven't hooked up the scanned yet, but that is the next thing. For now I'll share the FireWire with my camera since I don't use them at the same time.
Don't know what I'm going to use the eSATA port on the front for or for that matter the Thunderbolt port on the back.
I'm laying out my "studio" now and liked your setup on Oklahoma. Any guidance for counter heighth? Do you use your keyboard on the counter or from a slide out tray?
If you don't need all the ports on the system, that's ok - they were not extra cost, and are there if you do ever have the need. Not sure what to say about noise - the system itself (fans) should be very quiet, but sounds like you have a noisy hard drive there.
Never built a custom desk and have always just put the keyboard on the desk, and I then have a place to rest my forearms while I type. I think most desks are around 29-30" high on average.
About DPI, let's see if I can explain in a way that makes sense. I'll start with scanning. You have to tell the scanner what "precision" to use when converting the print image into a computer file. The higher the DPI, the more detail that will be preserved in the scanned file. More dots = more data.Once an image is "digitized" into the computer and you look at the file in Photoshop or other graphics programs, see will see that the image has dimensions measured in pixels (dots) on the X and Y axis.
Scanning a 4" x 6" photo at 300dpi results in an image with dimensions of 1200x1800 pixels. NTSC video is 720x480 pixels, 720p is 1280x720, and 1080 is 1920x1080. So depending on which video format you wish to edit/deliver, did your scan produce at least enough resolution to match or exceed the video format? If the scanned image size is smaller than the video requirements, your image would be scaled up in the edit software, which will result in a loss of quality. So if scanning a wallet-sized photo, or slides or film, you obviously need to use a much higher dpi so that you end up with enough pixels to match or exceed the video frame size, and of course if you intend to reprint the scan, then the more quality (dpi) the better! So you might want 1200 or 2400dpi scanning for smaller sources, while for an 8x10 source, 300dpi may be plenty.
It is ok (and preferable) for the still image to have a higher resolution than is needed for the video. Printing is one reason to scan at the highest quality. Another reason is that for video, a larger-than-needed still image affords the editor the ability to add zooms and pans to the still to bring it to life. Sure, you can add motion to a video-sized still, but the moment you start zooming in, you lose quality. If the still is larger than needed, you then have extra pixels to play with and can zoom in (to a degree) without quality loss! Of course, you don't want the image to be grossly oversized or it can bog down the edit system.
In that case, scan at the highest resolution so you have good quality for reprints, but then batch process the pics down to a more reasonable size for video, as suggested by Bill. So let's say you are going to produce an SD video at 720x480. Your scans are huge, maybe 5000x5000. If you knock them down to 1500x1500, it will be a lot easier on the edit software, but since the still image has twice the resolution as needed to fill the video frame, you then have the option to zoom in on a subject in the photo, up to 50% closer, before you would start to lose any quality. If you need to zoom in really tight, then make the image larger as needed. If you don't want to add any motion, then get the stills as close to the target video frame size as possible.
If I have a still image at 720x480 in my NTSC video project, any zooming degrades the image. If using the 5000x5000 source image, I can then zoom in on one person in a large group and still have crystal clarity in the video since the reframed portion of the larger image still has at least 720x480 pixels (or more) to offer, so nothing is being "blown up" yet, still using original source material. But as stated, if not zooming in to that degree, the extra pixels are wasted and will only serve to slow down the system, so scale the images down accordingly before importing into Premiere to prevent bloating of the project files.
Another little tip - if you intend to start with a larger image and animate a zoom in Premiere, you may encounter flickering of fine details in the video. If that happens, I like to just add a touch of Gaussian Blur in Photoshop, maybe 0.1 or 0.2 is enough. No one will ever notice the very slight blur, but the animation will then look 100% better, a night and day difference! Remember to blur a COPY of the image and not the original scan, since you would not want blur added to something you might later print or repurpose.
Have fun with your project
Safe Harbor Computers
I have published a series of articles on the subject at the links below which you will find helpful.
The short story is that so much image quality is lost during the conversion to video, even HD, that original resolution doesn't really matter.
Thanks for chiming in. The link you sent was primarily about color, with a short blurb at the end which basically said "video sucks anyway so it doesn't matter what the source image is" which I think is kind of misleading. Your other article about resolution is much more pertinent to this discussion and is a good resource, thanks - http://dpsdave.hubpages.com/hub/What-makes-a-high-quality-photograph-scan
I totally understand what you are saying - if I take a 10 megapixel photo with a new Canon DSLR, even a good HD video will never be able to reproduce the full quality, I get it. However, if scanning, which is what we were discussing, then the scan quality (image resolution) definitely impacts the video quality.
If I'm scanning a tiny wallet photo at 150 dpi or even 300 or 600dpi, I won't even have enough resolution to fill an SD video frame, let alone HD. So the image source quality does indeed matter when transferring to video, don't you agree?
Also, I can't argue that the NTSC color system was weak, but that only applies to standard definition videos, as HD video is all digital and uses REC. 709 color. It's not computer color, but certainly much better than NTSC ever could be.
Thanks again for your contribution to this thread, I'm learning more every day
David R. Orr
President, Digital Photographic Service
DpsDave Mail Logo
I've got some 3000 slides to scan and while .tif has been around for some time I was hoping to be able to get some compression to save on file space.
There's nothing to stop you from considering TIFF, because you have several options for compressing it without loss of quality. However, there's a lot of traditional advice being given out and some of it needs adjusting for today's workflows.
TIFF vs PSD: The TIFF format can store just about anything PSD format can (layers, Smart Objects, effects, etc.), and it's more open, so using TIFF can potentially let you open the files in the future using many applications other than Photoshop or Lightroom.
TIFF compression: When you Save As a TIFF in Photoshop (or export TIFF in Lightroom), you get some compression options. The traditional advice is to use LZW compression. That was excellent advice for the 8-bit age, but now that 16 bit files are both recommended and common there are better ways. Adobe provides a ZIP compression option for TIFF and that compresses even more than LZW. There is another very serious reason not to use LZW if you work with 16-bit files: In some cases applying LZW to a 16-bit TIFF can actually make it much larger!
An example I just tried with a 16 bits/channel TIFF:
Uncompressed - 107.5MB
Compressed with LZW - 121.3MB
Compressed with ZIP - 96.2MB
The disadvantage of ZIP is it takes the longest to open and save. If you're working on an image a lot you might temporarily use PSD. PSD is actually compressed, but you don't have any control over it.
I know this question has already been answered a lot, according to the use of scanned film, but thought I'd add some more for fun.
DPI ( dots per inch ) is from the days of print and offset press originally. It is an intaglio method of capturing ink in little depressions of a plate so that an image can be reproduced using dots of ink in the plate. The more dots ( depressions) the more resolution. 72 dpi is what newspapers use to print the newspaper ( you can see the dots with a magnifying glass for pictures on newspaper ).
300 dpi is more smooth looking.. more dots, more ink in smaller space ( inch ) and more clarity.
Some film scanners ( google nikon coolscan ) use the dpi as a measurement of the 'clarity' of the optical scan. The output is then into the digital realm and is now PPI ( pixels per inch ).
Our computer monitors typically have only 72 -90 pixels per inch that they can display. So our computer monitors are NOT as clear and fine as an offset press ( on paper ) " print" of the same image. To make things a little more complicated the video stuff is typically 72 ppi ( cause TV's and monitors , lcd's etc are not up to par with the clarity and tight grouping of those little dots that historically have been available on paper yet ).
I know this sounds weird, but if you look at the specs for my earlier post about scanner and do google search re: nikon coolscan you can maybe get the general idea. In the beginning was dots for ink, now we have pixels.
I hope this helps because frankly if I had 1000 slides or film stuff ( 35mm and 120 etc ) I would want to sorta get this fundamental info down pat.... cause that's a lot of stuff to scan.
If you want the best quality for both print and video you have to sorta maybe think about the high end first ( print ) and then you can downscale stuff to go video land. One ( print ) is 300 ppi ( see my earlier post re: output for minolta dimage scanner ).. and the other is 72 ppi ( video ).
The ppi and dimensions in inches is reciprocal ( you lessen ppi, you gain dimension in inches )....keeping the file size the same. Good luck !
edit: whoops , sorry, I woke up last night in middle of night and realized newspapers are more like 100- 120 dpi.. you can see the dots..thats where the ink is...but its not 72...thats video monitors ...sorry...