![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
vectorizing
- Thread starter badjoke
- Start date
Are you talking about making vector art? Here's a cool tutorial I found.
http://www.heathrowe.com/tuts/vector.asp
http://www.heathrowe.com/tuts/vector.asp
The Corel suite includes a tool called CorelTrace ... Flash also
offers tracing functionality ... although I find it tipically does not
work as well as I would like, I therefore import the raster image
into CorelDraw (vector illustration software) and trace it by hand.
But then ... I am anal!
offers tracing functionality ... although I find it tipically does not
work as well as I would like, I therefore import the raster image
into CorelDraw (vector illustration software) and trace it by hand.
But then ... I am anal!
NightWolfe
Gawd
- Joined
- Sep 30, 2002
- Messages
- 786
Actually Vectorizing is turning a Raster image into a Vector Image ..
and if you want the quality to remain the same the Vector image is going to be at least 100 MBs ..
Actually vectoring an image REDUCES its size...
You can create an extremely complex vector image and it will still come in under 3 MB... Rasterize it at 300 PPI and it will come in at around 50 MB.
Vectoring is nothing more than mathematically equating lines and shapes, as opposed to raster which is drawing EVERY single pixel on the image.
But back to the original question, if you want to vectorize an image, d/l or buy Adobe Streamline, it specializes in just what you're looking for.
You can create an extremely complex vector image and it will still come in under 3 MB... Rasterize it at 300 PPI and it will come in at around 50 MB.
Vectoring is nothing more than mathematically equating lines and shapes, as opposed to raster which is drawing EVERY single pixel on the image.
But back to the original question, if you want to vectorize an image, d/l or buy Adobe Streamline, it specializes in just what you're looking for.
chevy4x4trucks
Weaksauce
- Joined
- Nov 6, 2003
- Messages
- 85
I'll second that. I work w/ AutoCAD all day which uses vector "images" if you want to call them that. I can have a complex layout of lets say a baseball satdium and the file size still comes in under a megabyte.
Also, working in flash (which also employs vector graphics) the layout of my portfolio site together with the animation weighs in at a whopping 8kilobytes which is about what a 40x40 jpg at medium resolution would be. Not bad.
Vector images are the best way to go becuase you can crank out the highest quality at any size.
As for the original topic of the thread. Flash has a "trace bitmap" feature which works decently. Otherwise the way I go about it is tracing by hand in either illustrator of flash or even autocad then importing into illustrator.
All of you guys are right. Vector images usually take up much less space than bitmap images, but if you were to take a bitmap image and convert it to vector such that there was no loss in quality it would probably be pretty complex and probably end up being larger than the bitmap. However, if you were to take a bitmap image and vectorize it to look more like a line art image (aka Badjoke's guitar player above) then yes it will be much, much smaller.
Think of it this way, in order to make a vector image of say a photograph and not lose any quality, the program would theoretically have to map out a vector for every pixel in the image. You'd basically end up with an image containing a bunch of tiny vectorized squares (or very close to that anyway). That means a mathmatical equasion AND an 8 or 16 bit color assignment for every vectorized square rather than just an 8 or 16 bit number to represent a color for each pixel. In this case the vector image would be larger. (this also doesn't take into account any file compression on the bitmap file, such as jpeg).
Now, if you don't care about retaining the exact quality of the photo and you want it to look like a line art drawing (aka badjokes guitar player again) then not every pixel in the image needs to turned into a vectorized shape. The pixels are grouped together in much larger "generalized" areas of color and made into a vectorized shape, thus giving it that line-art or illustration quality. In this case the vector image would be much smaller.
As to the original question, I'm going to go with Rocketpig on this one and say Adobe Streamline is what you're looking for.
Think of it this way, in order to make a vector image of say a photograph and not lose any quality, the program would theoretically have to map out a vector for every pixel in the image. You'd basically end up with an image containing a bunch of tiny vectorized squares (or very close to that anyway). That means a mathmatical equasion AND an 8 or 16 bit color assignment for every vectorized square rather than just an 8 or 16 bit number to represent a color for each pixel. In this case the vector image would be larger. (this also doesn't take into account any file compression on the bitmap file, such as jpeg).
Now, if you don't care about retaining the exact quality of the photo and you want it to look like a line art drawing (aka badjokes guitar player again) then not every pixel in the image needs to turned into a vectorized shape. The pixels are grouped together in much larger "generalized" areas of color and made into a vectorized shape, thus giving it that line-art or illustration quality. In this case the vector image would be much smaller.
As to the original question, I'm going to go with Rocketpig on this one and say Adobe Streamline is what you're looking for.
TwoHardCore
Weaksauce
- Joined
- Sep 28, 2001
- Messages
- 114
actually 3ee Dee answered badjokes question with a tutorial that explained exactly how it would be done with PS CS.
I don't think much more can be said about how it's done, as it's all there in the tutorial. PS is not a good program for creating vector art, a few vector paths maybe, but nothing complicated.
THC
I don't think much more can be said about how it's done, as it's all there in the tutorial. PS is not a good program for creating vector art, a few vector paths maybe, but nothing complicated.
THC
One
Raster images store a certain amount of colour bits per picture
element (fondly known as pixels), the size of the file is directly
proportional to it's dimensions. If compression is omitted,
complexity has no effect on the file size, but in the case of non-
lossy compression image compexity has visible adverse affects on
compression efficiency, in the case of lossy compression the
negative impact on filesize is less obvious but the phenomenon is
still present.
Two
2D Vector illustrations use vector data to define lines and shapes.
Curves are typically represented as splines and therefore require
four two-dimensional vectors. Shading and other parameters for
surfaces can be stored parametrically, and are therefore
negligable in size.
Three
Tracing software locates similar hued areas within a raster image,
and then constructs suitable vector delineations around the
homogenous chromatic area. Obviously if the tracing software is
instructed to maintain perfect image accuracy every pixel would be
traced as a square (this would be a worst case of complexity and
tracing efficacy).
Four
In the worst case (let us assume a black and white chequered
pixel pattern) an uncompressed raster image of 1000px x 1000px
x 32bpp would amount to roughly 3.8MB (sans header info)
In the worst vector case, every corner of a square would be
saved in two dimensional floating point values thus 4 x 2 x 64-bit,
which means roughly 64MB (sans header info)
Five
badjoke, if the image you intend to enlarge is a line drawing type
raster image, give it to me and I'll do it for you. If however the
image is a photgraph of you felching a gerbil, it would be
impossible to scale it up without some image loss since image
data can only be interpolated and that is never accurate. The
best interpolation algorithms can only guess at pixel values
between the ones you already have.
Raster images store a certain amount of colour bits per picture
element (fondly known as pixels), the size of the file is directly
proportional to it's dimensions. If compression is omitted,
complexity has no effect on the file size, but in the case of non-
lossy compression image compexity has visible adverse affects on
compression efficiency, in the case of lossy compression the
negative impact on filesize is less obvious but the phenomenon is
still present.
Two
2D Vector illustrations use vector data to define lines and shapes.
Curves are typically represented as splines and therefore require
four two-dimensional vectors. Shading and other parameters for
surfaces can be stored parametrically, and are therefore
negligable in size.
Three
Tracing software locates similar hued areas within a raster image,
and then constructs suitable vector delineations around the
homogenous chromatic area. Obviously if the tracing software is
instructed to maintain perfect image accuracy every pixel would be
traced as a square (this would be a worst case of complexity and
tracing efficacy).
Four
In the worst case (let us assume a black and white chequered
pixel pattern) an uncompressed raster image of 1000px x 1000px
x 32bpp would amount to roughly 3.8MB (sans header info)
In the worst vector case, every corner of a square would be
saved in two dimensional floating point values thus 4 x 2 x 64-bit,
which means roughly 64MB (sans header info)
Five
badjoke, if the image you intend to enlarge is a line drawing type
raster image, give it to me and I'll do it for you. If however the
image is a photgraph of you felching a gerbil, it would be
impossible to scale it up without some image loss since image
data can only be interpolated and that is never accurate. The
best interpolation algorithms can only guess at pixel values
between the ones you already have.
I doubt whether such a filter exists.
<aside>
Why in hell do PS whores believe there's a PS filter to unblock
your toilet, or to was your car, or to blow your nose, or to make
coffee, or to find a sexy babe? Sometimes you have to actually
roll up your sleeves and do art!
</aside>
Besides ... it's vectorising, thus it will take a raster image and
convert it to vectors ... this is not typically in the feature-domain
of PS anymore, unless of course you reconvert it to raster, but no
filter that I know of will do that kind of thing.
<aside>
Why in hell do PS whores believe there's a PS filter to unblock
your toilet, or to was your car, or to blow your nose, or to make
coffee, or to find a sexy babe? Sometimes you have to actually
roll up your sleeves and do art!
</aside>
Besides ... it's vectorising, thus it will take a raster image and
convert it to vectors ... this is not typically in the feature-domain
of PS anymore, unless of course you reconvert it to raster, but no
filter that I know of will do that kind of thing.
Lord Hyperion, thank you.
Sometimes you just have to do ART or HAVE TALENT.
I'm not a snob but the biggest thing killing my profession is jackasses thinking they can do a logo, website, brochure, or whatever for 1/4 the price I charge.
Only problem is that their material is so pathetic that it hurts me to see that anyone paid for it, much less allowed it into mass-circulated print.
Sigh...
Sometimes you just have to do ART or HAVE TALENT.
I'm not a snob but the biggest thing killing my profession is jackasses thinking they can do a logo, website, brochure, or whatever for 1/4 the price I charge.
Only problem is that their material is so pathetic that it hurts me to see that anyone paid for it, much less allowed it into mass-circulated print.
Sigh...
-----
I doubt whether such a filter exists.
<aside>
Why in hell do PS whores believe there's a PS filter to unblock
your toilet, or to was your car, or to blow your nose, or to make
coffee, or to find a sexy babe? Sometimes you have to actually
roll up your sleeves and do art!
</aside>
-----
I didn't say there was a filter that was create a vector from a rastered image, I said there was a filter that could take small pics and create larger ones with similar quality.
http://www.lizardtech.com/solutions/gf/
Cads
I doubt whether such a filter exists.
<aside>
Why in hell do PS whores believe there's a PS filter to unblock
your toilet, or to was your car, or to blow your nose, or to make
coffee, or to find a sexy babe? Sometimes you have to actually
roll up your sleeves and do art!
</aside>
-----
I didn't say there was a filter that was create a vector from a rastered image, I said there was a filter that could take small pics and create larger ones with similar quality.
http://www.lizardtech.com/solutions/gf/
Cads
I checked out the link ... and if I had to wager a guess on how
that plugin works it would be as follows:
It does a field analysis of the image data (in all possible color
dimensions) and encodes any 'ordered patterns' discovered by
fractal geometry. Should you scale the image up, you have a
mathematical model (and therefore an inaccurate approximation)
of the original.
This method will perhaps allow for better enlargement than say
bi-linear, bi-quadratic, or even bi-cubic interpolation, HOWEVER,
it is not perfect. As I have said before, you CANNOT create data
that does not exist. Look at this quote from their site:
"Resolution on demand lets you output any file size from a single
encoded file, up to 600% without image degradation"
So there is a limit on the correct functioning of the plug-in, and
even in the 100% - 600% range the scale up, will not be PERFECT.
So claiming resolution independence as they do, is not entirely
honest.
If you don't believe me, shell out the $159, and try this:
1. Open a 1024px x 768px x 32bpp photograph.
2. Select the entire image and copy it to the clipboard.
3. Downsample it to 200px x 150px x 32bpp.
4. Scale it back to 1024px x 768px x 32bpp with your 'miracle tool'.
5. Paste a copy of the original image over the processed one.
6. Set the pasted layer blend mode to difference (this calculates
the difference in colour values).
7. Flatten your new image
8. Set your Brightness and Contrast up by 100%.
If you at this stage see anything but a 50% grey background ...
your 'miracle tool' is not 100% accurate.
that plugin works it would be as follows:
It does a field analysis of the image data (in all possible color
dimensions) and encodes any 'ordered patterns' discovered by
fractal geometry. Should you scale the image up, you have a
mathematical model (and therefore an inaccurate approximation)
of the original.
This method will perhaps allow for better enlargement than say
bi-linear, bi-quadratic, or even bi-cubic interpolation, HOWEVER,
it is not perfect. As I have said before, you CANNOT create data
that does not exist. Look at this quote from their site:
"Resolution on demand lets you output any file size from a single
encoded file, up to 600% without image degradation"
So there is a limit on the correct functioning of the plug-in, and
even in the 100% - 600% range the scale up, will not be PERFECT.
So claiming resolution independence as they do, is not entirely
honest.
If you don't believe me, shell out the $159, and try this:
1. Open a 1024px x 768px x 32bpp photograph.
2. Select the entire image and copy it to the clipboard.
3. Downsample it to 200px x 150px x 32bpp.
4. Scale it back to 1024px x 768px x 32bpp with your 'miracle tool'.
5. Paste a copy of the original image over the processed one.
6. Set the pasted layer blend mode to difference (this calculates
the difference in colour values).
7. Flatten your new image
8. Set your Brightness and Contrast up by 100%.
If you at this stage see anything but a 50% grey background ...
your 'miracle tool' is not 100% accurate.
That link does however take asmall pic and make it big with same quality. Commonsense should tell you that he didn't mean that it had to be 100% accurate.
So claiming resolution independence as they do, is not entirely honest.
I think they claim that because visually they are accurate, again the whole commonsense thing.
your 'miracle tool' is not 100% accurate.
Nothing is 100% accurate.
So claiming resolution independence as they do, is not entirely honest.
I think they claim that because visually they are accurate, again the whole commonsense thing.
your 'miracle tool' is not 100% accurate.
Nothing is 100% accurate.
Cadsworth, common sense is a dangerous tool when arguing.
"Take a small image and make it big with same quality"
That is a false proposition. Saying that would be like saying 145
is the same in magnitude as 147 ... they are close but not equal.
People of less technical orientation would for example believe that
this tool actually works completely, when in actual fact it is going
to let them down when working with masks. If you lose even the
slightest bit of accuracy on a mask your selection quality
deteriorates. Which brings me to your next false proposition:
"Nothing is 100% accurate"
Not true, there are transformations that are invertable with 100%
accuracy, this is especially true for operations in colour-space, for
example tone-curve changes can be inverted 100% correctly.
However, blur, sharpen, resample and in fact most other filters
permanently corrupt image data.
I am sure this tool does a better job of maintaining a lower margin
of error than other algorithms mentioned previously, yet, it is
still not perfect, and should be applied with care.
"Take a small image and make it big with same quality"
That is a false proposition. Saying that would be like saying 145
is the same in magnitude as 147 ... they are close but not equal.
People of less technical orientation would for example believe that
this tool actually works completely, when in actual fact it is going
to let them down when working with masks. If you lose even the
slightest bit of accuracy on a mask your selection quality
deteriorates. Which brings me to your next false proposition:
"Nothing is 100% accurate"
Not true, there are transformations that are invertable with 100%
accuracy, this is especially true for operations in colour-space, for
example tone-curve changes can be inverted 100% correctly.
However, blur, sharpen, resample and in fact most other filters
permanently corrupt image data.
I am sure this tool does a better job of maintaining a lower margin
of error than other algorithms mentioned previously, yet, it is
still not perfect, and should be applied with care.