FFT and Golf Balls - Page 4

Duv · #91 06-15-2005, 12:03 AM

I have all of Chris's IP plugins showing up in Filters but what are they under in the Plugin's folder?

Dave

Cameraken · #92 06-15-2005, 12:44 PM

Duv
I’m not sure I understand your question.
I assume you have downloaded FoveaPro/FFT

You use
Filter > IP*Fourier > FFT(Forward)
This gives you the FFT

Then you use
Filter > IP*Fourier > Apply Filter and FFT(Inverse)
This gives you the Image back

Hope this helps

Ken

Duv · #93 06-15-2005, 08:41 PM

Sorry Ken. Mine was a terrible post. I have all the IP*thisandthat's in Filters and I wish to delete it all. But I can't seem to find the appropriate files 8bf in the Plugin Folder. I see nothing called Fovea, IP or anything that seems to relate.

Cheers
Dave

Cameraken · #94 06-16-2005, 01:47 PM

Uninstall FoveaProFFT

If anyone has installed the trial version of FoveaPro and now wishes to uninstall it

The problem is that there is no uninstall program. Not even for users who have purchased one version and need to uninstall it to upgrade to a newer version.
With hindsight we should have created a restore point before installation that we could now go back to. However with the lack of hindsight the uninstall must be done manually.

Here are my suggestions

You originally downloaded FP3.zip or FP3.sit
These extracted to
PC FoveaPro3 Installer.exe
Tutorial5.pdf
PC Readme.txt
So where ever you downloaded these files originally – these can be deleted.

Now for the installed filters
When you originally installed this it would have installed into a default folder something like this
c:\Program Files\Adobe\Photoshop 7.0\Plug-ins\Filters\

The big problem is that all the FoveaPro filters are now mixed up with your other filters. Fortunately all the filters are dated the same so if you sort the folder by date all the FoveaPro filters will be grouped together and can all be deleted.
The date on the Windows version is 14/4/2003 (Mac may be different)

So here is the Easy way for Windows XP users
Click Start > Search
In the All or Part of Name Box Type 8bf Then click Search
Once the search is complete click the date modified tab and all the FoveaPro filters will be grouped together with the date 14/4/2003
Highlight all of them and delete
That’s it.

If any of you are uneasy about deleting any other filters created on the same day here is a list of the FoveaPro filters that need deleting. There ate 171 of them listed alphabetically.

ABSDIFF.8BF
AdaptEq.8BF
ADD.8BF
AlignW2.8BF
alvlbrgt.8BF
alvlcolr.8BF
alvldark.8BF
AplyCLUT.8BF
ApplyMsk.8BF
ASTMGSZ.8BF
AutoCntr.8BF
BandPass.8BF
BiLevel.8BF
BLEND.8BF
BoolGen.8BF
Britness.8BF
c2g.8BF
calibrat.8BF
Centroid.8BF
ClrPlane.8BF
CMedian.8BF
Colocal.8BF
ColorFlt.8BF
ColorMap.8BF
ColrCube.8BF
colrvalu.8BF
Contour.8BF
Contrast.8BF
ConvolFT.8BF
Convolve.8BF
CountGrd.8BF
CountMrk.8BF
CRange.8BF
CrossCor.8BF
CSpace.8BF
Custom.8BF
Cutoff.8BF
CVE.8BF
darker.8BF
Deconvol.8BF
DeIntL.8bf
Dens_Cal.8BF
divide.8BF
DOG.8BF
DrwCycld.8BF
DrwLinGr.8BF
DrwPntGr.8BF
EdgePres.8BF
edm.8BF
EDMMorph.8BF
ElevProf.8BF
Entropy.8BF
Equalize.8BF
extremum.8BF
Extrude.8BF
FeatAND.8BF
FeatDist.8BF
FeatID.8BF
FeatPlot.8BF
Features.8BF
FillHole.8BF
FillRegn.8BF
Fit_BG.8BF
Flatten.8BF
Focus.8BF
ForwrdFT.8BF
FractalDim.8BF
FrameROI.8BF
FreiChen.8BF
FTProfil.8BF
Fusion.8BF
Gaussian.8BF
GenMask.8BF
Global.8BF
Haralck1.8BF
Haralck2.8BF
Harmonic.8BF
HillDale.8BF
Histogrm.8BF
Hit_Miss.8BF
Hough.8BF
HoughInv.8BF
HSIThrsh.8BF
HVLines.8BF
HybridMd.8BF
IntensP.8BF
InteractiveDeconv.8BF
InversFT.8BF
InvFilt.8BF
IsoLines.8BF
Isometric.8BF
ItMidDis.8BF
Kirsch.8BF
Kuwahara.8BF
Lighter.8BF
Limits.8BF
LineLen.8BF
LineProf.8BF
LocalEq.8BF
LogNewLine.8BF
MagBar.8BF
MakOpaq.8BF
Mask.8BF
MatchHue.8BF
Math.8BF
MeasIcpt.8BF
MeasPwSp.8BF
Median5.8BF
MorphGen.8BF
Multiply.8BF
NborDist.8BF
Norm_RGB.8BF
Olympic.8BF
Orient.8BF
Outline.8BF
Peaks.8BF
PerimvsA.8BF
Phase.8BF
Phong.8BF
Prune.8BF
PtIcept.8BF
PwrSpec.8BF
RadLines.8BF
RadProf.8BF
Random.8BF
range5x5.8BF
Rank.8BF
RankColr.8BF
RankGen.8BF
Recall2.8BF
Reconstr.8BF
Regions.8BF
Register.8BF
relief.8BF
Rem_BG.8BF
RollBall.8BF
Seed_Grw.8BF
Select.8BF
Setup_2.8BF
SetupLog.8BF
SetupRecipeFile.8BF
Shift.8BF
ShowFTVal.8BF
showval.8BF
SigmavsA.8BF
SineFunc.8BF
skel.8BF
SlitIsnd.8BF
SobelDir.8BF
SobelMag.8BF
Stereo.8BF
StochEdg.8BF
Subtract.8BF
SurfStat.8BF
Swap2.8BF
Symmetry.8BF
Tangent.8BF
texture.8BF
Thicken.8BF
Thinning.8BF
Thr_HSI.8BF
Thr_Pct.8BF
Thr_Skel.8BF
Thresh.8BF
ThrshPen.8BF
Tophat.8BF
Trim.8BF
TriStim.8BF
UEP.8BF
Var_Adj.8BF
Watershd.8BF

End of List

Ken

Duv · #95 06-16-2005, 08:22 PM

Thank you so much Ken for taking the time. I was getting a little anxious and concerned but everything is back to normal now.

Cheers
Dave

Cameraken · #96 06-23-2005, 01:51 PM

ByRo has now posted his tutorial on FFT RGB which can be found here

http://retouchpro.com/tutorials/?m=show&id=185

Thanks ByRo for a very clear Tutorial.

Mac users can follow along using image-J mentioned in this thread

Ken

deadants · #97 08-18-2005, 09:00 PM

Well I have had some interesting results paying with imageJ. It definitely works when you erase the centre star. If you make a small erase of the centre star the image is blurred. The bigger the erased circle, the sharper the image becomes. Be careful not to erase the other stars or the image turns black.

krishna · #98 08-19-2005, 10:54 AM

Its brilliant. This has made the process so easy and simple. I tried it with 3 images. Thanks for posting this trick.

--
Krishna

Cameraken · #99 08-19-2005, 01:10 PM

Image J is part of The Image Processing Toolkit
Available Here for Free
http://rsb.info.nih.gov/ij/download.html
ImageJ runs on Linux, Mac OS 9, Mac OS X and Windows
The Docs are here
http://rsb.info.nih.gov/ij/docs/
And the bit about FFT is here
http://rsb.info.nih.gov/ij/docs/menus/process.html#fft

Deadants
Thanks for moving threads. I used to be very confused about using FFT RGB and I thought talking about ImageJ in the FFT RGB Tutorial thread may confuse even more.
Maybe we need an ImageJ FFT Tutorial as well.

I quite like ImageJ. It’s easier to use than FFT RGB. The downside is that being a standalone program and not a plug-in you have to come out of PS to use it. However as this should be the first step it’s not that bad.

I am on a PC but the program should be the same on a mac.

OK now your problem.

Here are the steps.

Open ImageJ
File > Open and load the picture
Process > FFT > FFT
The FFT will now open as in attachment 2
Double click the colour Picker and choose Black as your colour
Double click on the brush and choose a brush size of 10 (on this picture)
Now paint out all the stars except the centre one as in the attached picture 3
Process > FFT > Inverse FFT
Save the results

And that’s it. Your picture should now be texture free and ready to load into PS.

There should be no need for a black layer as If done correctly the picture will not be transparent.

If you need any more help just ask and perhaps post your image a bit bigger.

Ken

deadants · #**100** 08-21-2005, 09:50 PM

Hi Ken,
Sorry about the delay in replying but it was a big weekend

I tried again as you suggest by crossing out the outer stars while leaving the center star in tact, but all I get is a black image. I have had good results with the erasing centre star technique. As the same people wrote ImageJ you would think it's got to work the same on both PC and Mac the same way. So I'm scratching my head.

Krishna, did you do the erase centre star technique or the black out the outer stars method? Also are you using a Mac?

Has anyone else with a Mac tried imageJ, if so which method are you using?

krishna · #**101** 08-21-2005, 11:03 PM

Quote:

Originally Posted by deadants

Krishna, did you do the erase centre star technique or the black out the outer stars method? Also are you using a Mac?

I used to do the black out outer stars method and it used to give good results. But your method is much easier and it gives better results (it is doing the blacking out in a different way). In the blacking out method, my experience has been that even after completing, there will be a few patterns left out. To remove them I used to slightly blur the image. But with this method the masking is so complete and the work is over in a few seconds.

Quote:

Originally Posted by deadants

Has anyone else with a Mac tried imageJ, if so which method are you using?

I have been using ImageJ regularly in Mac and have tried in PC also. In PC, I haven't found any difference except sometimes the system hangs. But that cannot be blamed on ImageJ alone.

--
thanks
Krishna.

Cameraken · #**102** 08-22-2005, 02:33 PM

Deadants

I tried your method of painting the centre star with white. Painting with white is adding to the FFT. This method does not remove the texture. What it does do is blur the image so that you can’t see the texture.

Painting in black removes from the FFT. If your image is going transparent then you must have removed too much.

Post your image and we can work through the steps;

Ken

deadants · #**103** 08-23-2005, 07:45 PM

Hi ken,
I wasn't filling the cenre with white, I actually deleted it. But I take your point that the image is being blurred. I have tried several times at blacking out the surronding stars in ImageJ on the Mac but the results are disapointing. The IFFT doesn't seem to do anything, it returns a picture that looks the same as the original.

Here is a link to a good sized halftoned image that we may be able to walk through the process.
http://tinyurl.com/dg3ph

krishna · #**104** 08-24-2005, 12:32 AM

I tried the following (see screengrab) and got a sufficiently neat picture. Mid grays get cleared up easily while the pattern on black are persistent.

--
Krishna

Cameraken · #**105** 08-24-2005, 03:16 PM

Deadants

I have downloaded your picture. It is an image made up entirely of black dots (well crosses) in the white areas and white dots in the dark areas.
If you remove the dots you will remove the entire picture.

FFT is a way of removing Texture. And seeing there is No texture in this photograph I would not even use FFT. However, after saying that the way Krishna has used your method is good.
Yes it is blurring the image but only a little and blurring would be the only way to fill in the dots using any other method.

In photoshop you can achieve the same results with Gaussian blur set at 2 (setting at 3 eliminates even more of the dots but is blurring more)

In my attachment I changed the mode from Indexed to RGB
I used Gaussian blur set at 2 on the face and Gaussian blur set at 3 on the rest.
I think this is slightly better than using ImageJ.

I’ve tried using the same method in FFT RGB and it does not work. So this is specific to ImageJ.

Ken

krishna · #**106** 08-25-2005, 02:53 AM

Quote:

Originally Posted by Cameraken

In my attachment I changed the mode from Indexed to RGB
I used Gaussian blur set at 2 on the face and Gaussian blur set at 3 on the rest.
I think this is slightly better than using ImageJ.

I’ve tried using the same method in FFT RGB and it does not work. So this is specific to ImageJ.

In this case what you say is correct, Ken. But I have come across certain other instances where photos with halftone dots will have to be blurred considerably to get an acceptable picture whereas to the result you get using FFT in ImageJ, you need to add only a slight blur to remove vestiges of pattern.

As you say correctly, both the methods are blurring the image, but may be the algorithms being used (or the methods) might be different and hence the different result.

--
Krishna

Cameraken · #**107** 08-25-2005, 01:49 PM

Krishna

I am not very familiar with halftoned images. But I don’t think Deadants picture was very representative of what we normally see. You may be right about the way imageJ may blur differently, I don’t know.

If you have a picture where you think imageJ has worked better then please post it.

Ken

krishna · #**108** 08-26-2005, 02:50 AM

Here is one sample Ken. You can downloade it from
http://www.ola.in/tmp/photo2-sample.jpg

After FFT processing the picutre has been blurred slightly.

--
Krishna

PS: This is a picture scanned from a magazine.

Cameraken · #**109** 08-26-2005, 03:57 PM

Krishna

I’ve downloaded your sample Krishna.
Have you scanned this in?
If you have then are you aware of the de-screening options built into the scanner software. I would have thought that you could have got a better scan than this from a magazine.
Try scanning at a much higher resolution than you need, then drop the resolution in PS.
Also try scanning halftones at a 30 degree angle and then straighten up is PS. This can also help.

Anyway, working with what we have got.

This is a halftoned image with a slight texture from the paper it was scanned
The texture could be removed with FFT but I really don’t think it would help much as by far the biggest problem is the well spaced halftone dots.

A halftoned image has information missing which is lost forever. Filling in the dots can only be a guess based on the surrounding pixels. So whatever method you use the detail can not be increased.

Here is what I did

Opened the Sample image and cropped it back to the original
Changed mode to RGB
Duplicated the background twice
On BG Copy2 (the top one) Filter > Other > Offset Horizontal=2 OK, Layer Opacity 50%
Select All > Copy Merged > Paste
L1 Filter > Noise > Dust and Scratches 3,3 OK, Layer Opacity 80%
I then used brightness/Contrast and upped the contrast +30
Done.

I suppose you could try Polaroid Dust and Scratches or Neat Image but I don’t think you will get much better results.

Krishna. Your results are similar to mine. So unless anyone knows a better way I think a rescan is the only way you will improve on this.

Ken

Kraellin · #**110** 08-27-2005, 01:15 PM

got rid of the 'golf balls', but still have scan lines. just started playing with this FFT stuff. just a bit fussy. one thing i found is, you dont have to split the channels. you can edit the 'stars' in the color image you get after running FFT. and this isnt the FFT RGB one either. run fft, edit the cross marks and lines except the main middle one and then run ifft on that. seems to work just fine.

Craig

creeduk · #**111** 01-29-2006, 10:37 PM

Quote:

Originally Posted by creeduk

Just wanted to say this indepth thread made me go back and look at FFT, I had come across this filter in the past but misunderstood it, I had done the classic FFT then IFFT and see no result (of course..doh!) now reading this and going through old threads I have got it working. I still find it easier to use HSL as it shows the issue up clearer I think, I had some good success carefully cloning the vertical and horizontal lines which got rid of small highlights in that area that caused problems in the images.

So thanks for thrashing this subject again, this time it stuck for me

Having revisited this issue I have a sneaking suspicion that despite (or because of) having both filters I may have been using the wrong one. Hence my HSL comment. The examples I have worked on show the same pattern on red channel using the rgb as is seen on the blue channel using HSL version. Slaps forehead repeatedly.

mistermonday · #**112** 01-31-2006, 05:52 PM

Well, this thread is so long it probably doesn't need another tail. However, I keep finding ways to get better results. On a duplicate of this image, I reduced the contrast and ran a Discrete Cosine Filter which did a much better job than any of the FFTs I had. I applied the output to a contrast enhanced version of the original and used a luminosity blend. A slight clean up in Neat Image. Finally a quick trip to LAB to attenuate the bright yellow stains.
Regards, Murray