MISC GUIDES
-> DVD cases from low res images
The general rule of
thumb when working with images intended for print, is they
must be large resolution. This guide will help fix most
images that are medium resolution (300x300 pixels or
greater!) and would look bad with standard resizing methods.
You may want to also look over and refer to the guide on creating
DVD cases in Photoshop. This has some tips and
instructions on acquiring images and creating full DVD cases
in Photoshop.
This guide uses Adobe
Photoshop 7/8/CS/CS2, and the Genuine
Fractals plug-in for Photoshop ($80 version) to resize,
plus any VirtualDub
with MPEG reading abilities to extract images. And
optionally, TMPGEnc
(freeware or Plus) for cleaning video/images.
Getting
started
How do you know if a image is low resolution or not?
On the computer screen, in order to view an entire high
resolution image, you'll have to zoom out at least 15-50% or
more. In other words, anything that fits on your screen at
100% is probably too small. The only exception would be if
you have a mega-huge monitor, or have a resolution that
makes everything on your screen extremely tiny.
Examples of low-res images: (1) Video clips, even the
"high resolution" 720x480 DV source files, are way
too small. They need to be at least 200-300% larger. (2)
Most images found on web pages, again, way too small, even
ones that fill your screen. (3) Even many computer desktop
wallpapers are too small.
What is required for a high resolution image? This
depends largely on the quality of your printer and source.
There are really only two popular choices, 200dpi or 300dpi
print sizes.
- 200dpi print size uses a 1030
pixels wide by 1450 pixels tall canvas, for the
front and back of a standard DVD case.
- 300dpi print size uses a 1545
pixels wide by 2175 pixels tall canvas, for the
front and back of a standard DVD case.
Why
not just use a quick resize?
Enlarging an image in software will increase the size of
flaws. Digital compression schemes used for DVD-Video and
JPEG/GIF online images are very lossy, and errors are
usually not seen because they are small. When you increase
the image size, you increase the noise size too.
An original video tape was purchased used, in poor
condition, and without any case. The goal of this personal
project was to not only convert and restore the video to DVD,
but create a case too.
The following image is a medium-sized image. It comes from a
web site, it was a preview image for the VHS tape. Notice
how the errors can be seen in the solid color areas of the
helmets, and around objects. Looks carefully, they exist.
The image was then cropped to a DVD case size, for the front
of the case. This is how the image would look by simply
resizing to 1030x1450 resolution in Photoshop (shown here at
50% size, to fit on screen):
Notice all the nasty color noise in the white parts of the
uniform. It can be seen in the purple background too. The
helmet is especially horrible.
Now compare it to the version that was created using all the
steps you'll find below. Notice how clean it is, with the
noise gone, yet the detail left:
The only imperfection was the loss of the mostly-transparent
numbers on the top right side of the box. These were not
important anyway, so I will not fix them. Should somebody
REALLY need to fix that error, it would be simple to correct
in Photoshop (create a new layer, match the font, adjust the
transparency level, followed by a rasterize and delete brush
to erase excess, simple).
STEP ONE:
Obtain the image
For online images. As mentioned before, you may want
to also look over and refer to the guide on creating
DVD cases in Photoshop. This has some tips on acquiring
images.
For video clips. The easiest way to extract images
from a video clip is to use the free software VirtualDub.
Open VirtualDub and goto FILE -> OPEN VIDEO FILE, and
then open your video file.
NOTE! In order to use an MPEG file as source, you'll need a
version capable of reading MPEG/VOB files. In order to use
an AVI using a compression codec, the codec needs to be
installed on your system.
Use the slider bar at the bottom of the VirtualDub window
and move it around until you see a video frame you like. If
your video is interlaced, try to select a frame that does
not show interlacing lines. While you could de-interlace
later, it's not going to give optimal result. It would also
be wise to select an image that is low on noise. Maybe even
go for an image that is a close-up of something, and not a
busy image with tons of detail. You can both use the mouse
and the left/right arrow keys on your keyboard.
Once you've found the frame you want, go to VIDEO -> COPY
FRAME TO CLIPBOARD (pick either source or output, does not
matter).
Now open Photoshop, go to FILE -> NEW –
it should show the the resolution of the image in your
clipboard –
and press OK. Then go to EDIT -> PASTE. It should paste
your image into the window.
If your source was a video file, something like a 720x480 DV
file with a 4:3 aspect, then you may need to resize the
video file back to a 4:3 aspect, either 640x480 or 720x540.
This is a mild resize, you will not lose quality from it.
Aspect ratio is explained elsewhere on this site, in the
capture guides.
Now goto FILE -> SAVE AS as save the image as a TIFF file
for a backup.
If you plan to use STEP TWO, then we need to make
sure the image file has a resolution that is a multiple of 8
pixels. Pull out a calculator, or open the Calculator
software in Windows. Go to EDIT -> IMAGE SIZE and see
what the pixel size of you image is. If you used a video
file, it is already very likely a multiple of 8. If not,
change it.
The original image used in this guide was 288x475. The 288
was already a multiple of 8. The 475 was not. I entered 475
divided by 8 into the calculator, and got back 59.375. I
rounded up to 60 x 8 and got 480. See? Easy math. I resized
to 288x480, and was sure to disable the "constrain
proportions" option.
Now go to FILE -> SAVE AS and save as a BMP file.
STEP
TWO (optional): Clean the image
If your image is from the internet, or from a captured VHS
tape, or from an MPEG file, it could probably use some
cleaning. All of those sources have a bit of noise, and even
the minor problems, as was illustrated earlier, will become
a nuisance.
While a number of still image softwares have noise reduction
tools, for some reason the VIDEO NR filters in TMPGEnc can
outperform them by leaps and bounds, removing more noise,
and retaining more detail. The still image software tends to
just blur badly.
Open TMPGEnc. In this guide, TMPGEnc Plus 2.5 was
used. You can use either MPEG-1 or MPEG-2, should not
matter. Click BROWSE on the VIDEO SOURCE line, and select
your BMP image. If the BMP image is not listed as a file you
can open, change the filetype option to ALL FILES, and then
pick it.
Click the SETTINGS button. The file input and output
settings must match, for interlace, aspect and resolution.
Set the encode mode to either MPEG-1 or MPEG-2, does not
matter, and then give it a CBR bitrate of 15,000k. If using
MPEG-2, select non-interlace and DC of 9 (MPEG-1 does not
use these options).
Double-click on the the NOISE REDUCTION filter name, and it
will open the noise reduction options. Select the STILL
PICTURE filter, and adjust it as needed. For this example,
it looked best all the way to 100%. Adjust the pixel radius
as needed, 4 looked best for this example. Since this is a
single frame, the TIME AXIS filter does nothing, and is left
set to 0.
OK everything, then
START the encode. It will VERY QUICKLY create a new video
file for you. To extract the still from the video file,
simply repeat STEP 1, using VirtualDub to open the video and
then copy/paste the image into Photoshop.
Advanced options! If
your video file was pretty rough, and had a lot of noise,
then restore a small section of video with heavy filter work
in either TMPGEnc, VirtualDub or AVISynth prior to trying to
extract images. Use whatever is needed to remove nasty
noise, be it heavy temporal filters, chroma filters, or
other special filters. Then start all this. See the video
restoration section of the site.
STEP
THREE: Using Genuine Fractals and Photoshop
Now that the image has been extracted (and optionally
cleaned), it is in Photoshop and ready to use. At this time,
you can crop or do some very minor resizing (just a few
pixels), as needed. Then go to FILE -> AUTOMATE ->
GENUINE FRACTALS (or similar name, depending on the version
you have). If it requests you flatten the image, approve
it.
The GENUINE FRACTALS plug-in will launch, and you can resize
as needed. In this example, I resized to 200dpi and set the
height to 1450. The width was only 1013, but that's close
enough for this job, it will be resized to 1030 using
Photoshop resize.
There are some crop
options too, but these were not needed for this project, as
the image was cropped in advance. Personal choice on which
crop filters are used, GF or PS ones.
Genuine Fractals claims it can do a decent upsize job up to
700% on images. While that claim may be a bit exaggerated,
it does work quite well for the 400% range, as shown in this
example.
Conclusion
While some could argue the image is not perfect, or the
aspect has been changed by a few pixels one way or another,
or even the fact that it's been processed in so much
software, remember the source. What was once an unworkable
file is now a print-ready piece. While this would probably
never suffice for a major production (they'd use high res
custom art anyway!), it works quite well for personal
projects, small business needs, or even small run
productions.
Page Last Updated: December
24th 2005
|
Print
this guide:  |
|
