upsample/upscale a picture

[sinlg]

Upsample a pic.

http://actions.home.att.net/dSLR_Fractal_Sharpen.html

This is a cool action that works with Genuine Fractal to upsample a pic.

Anyone tried it with any success? Imagine a 12x zoomed pic enlarged somemore with this set of tools

[unseen]

Eh... i think you've mistaken this thing..
That link you provided is just crap

I've got this this proggie, definitely not some simple action.

the genuine fractal that you're looking for is here:
http://www.altamira-group.com/
this was formerly lizardtech

heh heh if you try hard enuff i know you can find a copy somewhere to use.
Works nice, but don't expect magic. This only performs wonders.

original


After downsizing


Upsizing using adobe photoshop bicubic interpolation


Upsizing using GF3.5



It's been rumoured that you can get acceptable enlargement by using 1% enlargement many many times (ok, you need to script an action for this)

[sinlg]

That dSLR link is supposed to be PS actions that makes use of Genuine Fractal to sharper an image? I guess it use GF to enlarge the pic, then do some functions for sharpening and downsizing the pic back to original size.

[unseen]

Originally Posted by sinlg That dSLR link is supposed to be PS actions that makes use of Genuine Fractal to sharper an image? I guess it use GF to enlarge the pic, then do some functions for sharpening and downsizing the pic back to original size.

oh.. heh sorry for being hasty.. saw wrongly.. :P
so sorry..

hmmm I use other software dedicated software for sharpening.
PhotoKit Sharpener rules.

[sinlg]

Everytime i open a photo in PS, it ask be about color space. If i choose discard, the pic looks right on the screen, if i say use Embedded colorspace, the colors looks off on the lcd monitor.

confused

[natnivek]

Originally Posted by unseen hmmm I use other software dedicated software for sharpening. PhotoKit Sharpener rules.

What is this? Share share leh....

[unseen]

Originally Posted by natnivek What is this? Share share leh....

PhotoKit Sharpener
heh that's the software.. hmmm but hope your computer's beefed up with ram.. or it'll slow your computer down...

[sinlg]

Was exchanging emails with the author of the dSLR sharpening action for Pshop using Genuine Fractal. He said he is in the process of releasing a version of the action that will be compatible with GF version 4.0

Would be interesting to see if it does the job any better (faster).

Also, was looking at Photomatix - an "exposure blending" or "contrast blending" software. looks interesting

[unseen]

Originally Posted by sinlg Also, was looking at Photomatix - an "exposure blending" or "contrast blending" software. looks interesting

Thanks.. this looks interesting..
will try it soon..

[sinlg]

Who can take 4 pics (with tripot) of a very dark scene with different exposure.. eg. 1 sec, 2 secs, 4 secs and 8 secs?

Can post here.. see if I can blend them together to form the ultimate "Exposure Blended" pic

[natnivek]

Originally Posted by unseen PhotoKit Sharpener heh that's the software.. hmmm but hope your computer's beefed up with ram.. or it'll slow your computer down...

Is this freeware? Sorry for the OT...

[sinlg]

why cant we just convert a typical picture into vectors first, and then enlarge, then rasterize them back to pixel form?

Wouldnt that be the ultimate enlarging solution? Would Adobe Streamline solve this problem?

[unseen]

Originally Posted by sinlg why cant we just convert a typical picture into vectors first, and then enlarge, then rasterize them back to pixel form? Wouldnt that be the ultimate enlarging solution? Would Adobe Streamline solve this problem?

heh you want the technical reply or you want the layman's reply? let's try the technical:

when you use a vectorised photo, you'll end up creating a few million vectors (imagine 5MP photo, that's 5million pixels, equating to 5million equations). Not only will it slow your computer down to a crawl, you might run out of memory space.

suppose our photo is of 2 characters wide
Original cell colour data:
AD

you vectorize it, you'll get vectors A and D. you upsize it, you will get
rescaled vector graphics
AADD

however if you stick with raster graphics, PS can perform this thing called bilinear or trilinear interpolation, means they calculate what's supposed to fill in the space between pixels when you increase the space between them.

rescaled raster graphics
ABCD

that's a reason why digital zooms are not good/blur.. they calculate what's supposed to be there and then they put it in.. the algorithm for calculation in PS is much better than a cam, and has a more power CPU as compared to camera. The results will always be better..

[sinlg]

Originally Posted by unseen suppose our photo is of 2 characters wide Original cell colour data: AD you vectorize it, you'll get vectors A and D. you upsize it, you will get rescaled vector graphics AADD however if you stick with raster graphics, PS can perform this thing called bilinear or trilinear interpolation, means they calculate what's supposed to fill in the space between pixels when you increase the space between them. rescaled raster graphics ABCD that's a reason why digital zooms are not good/blur.. they calculate what's supposed to be there and then they put it in.. the algorithm for calculation in PS is much better than a cam, and has a more power CPU as compared to camera. The results will always be better..

ok.. so far the enlarging software out there just doesnt cut it at least not at 800% enlargement ... otherwise, no need 8mp cameras, just get 1mp and upscale 800% hee

[unseen]

heh heh no software can ever "cut it"
After all you can't create details which are just not there... no way anyw software can reproduce the fine details supposed to be inside

[sinlg]

Thats what interpolation is supposed to attempt to do.

the more "intelligent" the interpolation technique, the better recreation is possible.

Perhaps all those CSI, spy movies exaggerate the way they can sharpen or enlarge pics

[unseen]

Originally Posted by sinlg Thats what interpolation is supposed to attempt to do. the more "intelligent" the interpolation technique, the better recreation is possible. Perhaps all those CSI, spy movies exaggerate the way they can sharpen or enlarge pics

you watch too much of those things..

imagine:
of a dress of very fine/thin black and white strips...
1) if you take far away it will look grey. you interpolate grey and grey, you get grey
2) only if you zoom in optically or move up close, will you ever be able to get the fine black and white strips

if interpolation can ever do half of what the movies show, then governments won't need to build spyplanes anymore.. just take a snapshot of the earth then interpolate/zoom in to find the exact locations of osama

[sinlg]

Originally Posted by unseen you watch too much of those things.. imagine: of a dress of very fine/thin black and white strips... 1) if you take far away it will look grey. you interpolate grey and grey, you get grey 2) only if you zoom in optically or move up close, will you ever be able to get the fine black and white strips if interpolation can ever do half of what the movies show, then governments won't need to build spyplanes anymore.. just take a snapshot of the earth then interpolate/zoom in to find the exact locations of osama

hmm.. but it takes too much computing power compared to just simply optical-zoom to the target