|
| |||||||
| Image Help Got a problem image? Don't know where to begin? Upload images and ask our users what they think or if they can help |
 |
| | Thread Tools |
|
#61
05-29-2005, 01:15 PM
| ||||
| ||||
| HSL may have better precision than RGB Notice how the workflow keeps getting more complex? I suspect that the RGB encoded version is worse than the HSL encoded version. For one reason, he should have used R & B, not R & G because most of the Luminance information is also stored in G. He should also have done a logarithmic µ-law encoding on Luminance, because that way you'd see more of the rays coming out from the stars and be able to do more accurate correction. Low intensity values will be overrepresented and high intensity values will have big truncation errors. #1 - Lobby him for a 16-bit version of the plugin. This will solve most of your woes. #2 - If your stars are white on a black field, you really need to erase to black and perform the anti-aliasing (feathering on a selection or a soft black brush). Yes, I know all about Image-J. Your tax dollars at work.  Prior to that, Wayne Rasband was working on NIH-Image, which had a bad combination of poor science and being free (he's a government employee). Thus, people used it and had no reason to pay for a *real* package that had the requirement of using good science. It also made it really difficult to compete in the early 90's for the low end of the market. Image-J has just been a sport since then. If you can fit it into you workflow, go for it. It is worth exactly what you paid for it. (If you don't consider that your tax money has been spent driving little guys like me out of business.)</rant>FFTW is just another implementation. Whose implementation is used is largely irrelevant as far as the quality of the result is concerned.  |
|
#62
05-29-2005, 05:34 PM
| ||||
| ||||
| Chirakov x Fovea (Round 1) If you look at the attachments first, you'll think that the first should be in Doug's Cloud thread and the second is a very uninspired art work.. read on... My final year project tutor taught me that when you want to discover something, devise an experiment - so this is what I did. 1) Take an "average" photo (Dunno what is average, but the golf balls ain' t); 2) Separate out Luminosity; 3) Expand canvas to some size multiple of two, fill in with 50% grey ; 4) Run through FFT, then imediately return through inverse FFT; 5) Crop back to the original size; 6) Compare with the original. The noisy image posted is an amplified (Levels 0 - 8) picture of the difference using the Alex Chirakov filter (average difference 3.5), the clean "art" is the Fovea filter (average difference 0.3). There was no significant difference between the RGB and HSB flavours of the Chirakov filter. Conclusions: -The Chirakov version does introduce a considerable amount of noise, whereas the Fovea process does not introduce any. (There is a 1 pixel difference, but this is uniform and to me that's not noise). - There's no difference between the two Chirakov filters, so you can use the easier (RGB) one without problems. Rô PS Chris - keep it coloured, otherwise the folks will think we're twins! |
|
#66
05-30-2005, 10:06 AM
| ||||
| ||||
| Wow! That’s heavy stuff. I feel like I need the beach after trying to understand that lot. I am not a professional retoucher. I don’t have hundreds of textures to remove. I hit this problem once or twice a year and just want to get the best I can from an odd picture. Byro Thanks for doing that comparison I think you are saying FFT RGB is as good as FFT HSL But FFT Fovea is Better. But if I’ve read correctly you are comparing the loss between the transfer What interests me is the end results and how to achieve it. So what probably interests me more is “What do I do to the FFT to improve the Picture?” The initial picture I supplied is a very poor example to use as a ‘Test picture’ What do you think of jcr6 cat picture – is that a good test picture? Jcr6 I’ve read chapter 4 and pages 9 and 10 of the pdf. Filter > IP*Fourier > FFT(Forward) Filter > IP*Fourier > FFT(Inverse) OR Apply Filter and FFTInverse The picture comes back different (darker the grey padding goes black) Could you please explain this? Also could you please clarify that FoveaPro should be working on a Luminocity Image? And if so then do you agree that byros method is OK. 1) Duplicate original image, right click on title bar and click Duplicate (that's a new image, not just a layer): 2) Set background colour to grey 127,127,127 (for Padding) 3) Image > Canvas Size > 512,512 in Pixels (In this case - square with dimension equal to an exact power of two) 4) On the new image, get Luminosity (New Layer (<Ctrl><J>), Edit>Fill - 50% grey, color, 100%.); 5) Flatten all, <Alt><L><F> (FFT does not understand layers); 6) Filter > IP*Fourier > FFT(Forward) 7) Paint out the Stars in white with a soft brush 8) Filter > IP*Fourier > Apply Filter and FFTInverse Does that sound OK? My suggestion earlier about adjusting brightness and contrast to ‘see the stars easier’ This does not seem to spoil the final image. What do you think? Flora. You did ‘cheat’ a little. You added the extra step. First I 'healed' out the stars in the Blue Channel (figure1), then in the Green Channel (figure 2) and finally I 'cleaned' the 'composite' RGB (figure 3). 'cleaned' the 'composite' RGB (figure 3). Is not in the manual http://www.retouchpro.com/forums/sho...59&postcount=6 Perhaps you could explain this step then we can find out how to apply. Ken |
|
#67
05-30-2005, 05:49 PM
| ||||
| ||||
| Hi everybody ... Quote:
  .... considered taking up knitting as a hobby ... Quote:
 I know the 'mathematicians' here would gladly burn me at the stake after this, but by 'cleaning' the composite RGB I meant just that .... after cloning/healing the stars from the two 'recomended' Channels, the composite RGB still looked a bit 'untidy' ... so .... I cleaned it before performing the IFFT...   |
|
#69
05-31-2005, 12:08 PM
| ||||
| ||||
| 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  |
|
#70
06-01-2005, 04:46 AM
| ||||
| ||||
| Hi Pete, Quote:
Quote:
It's not much... but I hope this helps.... |
|
#72
06-02-2005, 05:31 PM
| ||||
| ||||
| Tutorial? Despite the shortcomings of the filter mentioned, I wonder would Cameraken, Flora, or someone else here like to put all this together as a tutorial in the tutorial section? I know that the steps are listed out in several threads already, but having it all in one spot would be good - and I thought the images in Camerakens step by step were very clear and would help someone coming to the process new. In fact I think just taking those posts and putting them in the tut section would be great. (As an aside I tried using Ro's step by step previously, but without screenshots it mean I became very frustrated when it didn't work. It was only a month later when I tried it on another image I realised it was because the photo I was working on did not have a suitable pattern therefore did not have any 'stars'.) |
|
#73
06-02-2005, 06:30 PM
| ||||
| ||||
| And just to add to caitlen's suggestion, it would be great if their was a Macintosh tutorial as well. I did see that there are a couple of packages for the Mac, but I'm not sure if there are any tutorials on those. |
|
#75
06-02-2005, 08:51 PM
| ||||
| ||||
| Thanks byRo, I look forward to being able to get to grips with this technique. I may only see a couple of photos a year that fall into this category but when you have them sitting on your monitor staring at you, it's good to know there is a solution. The following attachment was presented to me yesterday. I don't know if it's a good candidate for FFT or not but it would be worth a try. |
|
#76
06-02-2005, 09:09 PM
| ||||
| ||||
| Thanks for your comments Caitlin. I started this thread because I was having so much trouble, especially with the newer FFT RGB for which the tutorials are very scarce. I wrote that tutorial to help everyone else struggling with this new version I have learnt a lot from this thread and I thank everyone so much for their contributions so far. As you mentioned FFT is not perfect for all images but as an initial starting point for removing texture IMO there is nothing to beat it. I hope there is more input regarding FoveaPro as this looks good. But I think we will have to wait for byRo to do this. I would be happy to re-write the tutorial and move it to the tutorial section. And if any mac users would like to volunteer their services we could perhaps do a version for deadants. But I don’t think there is a mac FFT Alex-Chirokov available. However as Byro has offered to do this we may be better leaving it to the experts. Off Topic The original picture I posted was from a much larger picture. By coincidence I have just been asked to do another so I thought I would show you where it came from. There may be a copyright issue with posting this so I’ve blurred it but it will give the general idea. This picture is 24” x 8” and the original photo I posted was from part of one of these – Hardly surprising that texture became a problem. I have restored a bit. few more hours to go. I use ExclaimPT Tutorial to Stitch the scans which works great. Ken |
|
#77
06-08-2005, 04:49 PM
| ||||
| ||||
| FFT tutorial For whoever is planning to do the FFT tutorial, I happened upon this one - Rather nicely done - a shame it's in Greek! FFT tutorial (Or in this case babelfish English) The original is here: http://www.dpgr.gr/index.php?page=FourierPS |
|
#78
06-09-2005, 03:00 PM
| ||||
| ||||
| Well found Caitlin. This is a tutorial on FFT RGB so it is the version we have been talking about here. Babelfish did a pretty good job of translating this. This is what I managed to understand from this site 1) Again another confirmation that the stars should be painted out in black 2) They are using a square picture as an example but do not seem to insist on square 3) They use blending mode set to colour to put the colour back. 4) They are using gaussian blur on the final image to clean up further. Does anyone speak Greek Ken |
|
#81
06-09-2005, 04:34 PM
| ||||
| ||||
| Quote:
Quote:
Quote:
Quote:
Caitlin, I got the hint - now I going to write it in portuguese, OK? Rô Yeah Duv, I know........ |
|
#82
06-10-2005, 04:45 AM
| |||
| |||
| Quote:
"Again another confirmation that the stars should be painted out in black" Well, actually he says that one should "mask-out, filter-out, delete or whatever" the white stars, claiming it's just a matter of terminology and does not actually make much of a difference... not very precise I'd say If anyone is interested, I could write an exact translation of that article, but I seriously doubt that it would shed any light to the subject. the posts here are much more precise and helpfull than anything I have found on there (and FFT is the one reason I keep coming back to retouchpro over and over again) Thnx... Jocker |
|
#83
06-10-2005, 05:56 AM
| ||||
| ||||
| Hi Jocker, Welcome to RP!  We do hope you'll keep coming back ... and, boy, are we glad you shed some light on this .... As you can see Rô had nearly finished translating the tutorial...even though, I think he was struggling a bit to stay as close as possible to the original text here... Quote:
Quote:
 Thank you very much for your help! |
|
#85
06-10-2005, 09:00 AM
| ||||
| ||||
| Welcome Jocker and thanks for your input. ‘Everything is just plain-old maths’ When you scan a picture it just becomes a set of numbers, which can be changed or manipulated. The Alex Chirakov FFT RGB is just a calculation on these numbers. The whole thing could be done in Excel or Visual Basic in fact I have a visual basic program for removing noise from sound, which uses the same calculation. The point about symmetry is a good one and makes sense with the maths. I don’t think it is necessary to translate the whole thing but can you help with the gaussian blur bit. It sounds a bit like Floras 'cleaned the composite' step. But as byRo says it sounds like they are talking about the colour layer. Ken |
|
#86
06-11-2005, 07:15 AM
| |||
| |||
| Quote:
---------------------------- Translating http://www.dpgr.gr/index.php?page=FourierPS ---------------------------- Let's see how a fourier transformation of our ewample looks. Since the information is pictured through the frequency spectrum, and the noise is periodical, like magic, all the noise is shown on specific places, much easier to process. (we use the filter Fourier Transform->FFT): [picture of fourier spectrum here] Doesnt it look odd that from this picture we can go back to the original? Notice how symmetric to the main star the bright points are. These are our noise. So what do we do? We delete, mask-out, filter-out, it's just a matter of terminology Using paintbrush we blank-out the aforementioned points, as in the picture bellow. The process was delibaretly done roughly so as to show that even this way, the method gives very good results. [picture of EDITED fourier spectrum here] Time to go back to the space domain from the frequency domain... (that's what a mathematician would say), we just apply the inverse filter and go back to the original picture: (Fourier Transform->IFFT) [grayscale transformed picture here] Look at how clean the picture looks. The fact that it's grayscale (that's how the filter FFT works, at first it transform it into grayscale and then applies the transformation), wont make it more difficult to us, it's enough that the luminocity is cleared. We take a copy of the original picture, put it over the final one as a new layer with blending mode set to colour and we apply as much gaussion blur needed to get rid of the noise (No detail is lost, as that is in the underlying layer). Here is the result: [final picture here] and a crop of before and after, blown up by 200%: [blown up crops here] Frequency domain transformation requires skills and experience. Good thing that there are various filters and utils that hide these dificulties The reason for this article was not to suggest that one should do this by hand, but to shed some light into how applying such methods can give very good results. --------------- End translation ----------------- See ya... Jocker Last edited by jocker; 06-12-2005 at 05:36 AM. |
|
#87
06-11-2005, 01:23 PM
| ||||
| ||||
| Thanks Jocker for taking the time to do that. Byro was right (as usual) :- ‘gaussion blur the colour Layer as much as needed to get rid of the noise No detail is lost, as that is in the underlying layer.’ That’s worth noting in case the info has come too late for byRo tutorial. Thanks again Jocker. Ken |
|
#88
06-12-2005, 02:21 AM
| ||||
| ||||
| jocker, This is just great!!! Thank you again for your time and help!!!!! Ken, ... about blurring the 'color' Layer .... it is right you don't lose details ... but, if to fade the noise you use the Gaussian Blur, careful because the colours tend to 'bleed' into each other so you can get 'noise-free' pink teeth or, if you are working on a portrait, the skin around the edges tends to get the background tinge ...etc. So, in this case, I would rather use the Median Filter as noise remover, as the Median Filter tends to 'recognize and respect' the edges between colours .... |
|
#89
06-12-2005, 05:43 AM
| |||
| |||
| Quote:
Jocker |
|
#90
06-12-2005, 02:03 PM
| ||||
| ||||
| Quote:
 I'd forgotten that. I've written the FFT tutorial but the "uploader" isn't cooperating with me this weekend, so I haven't been able to post it yet. Rô |
|
| Thread Tools | |
| |
Magic Bullet 
welcome to RetouchPRO!! 
Promise I'll post the tutorial this weekend, promise 
