|
Post by John Wall on Feb 26, 2012 23:38:20 GMT
Oh it wasnt just me thow!! Though......
|
|
|
Post by George D on Feb 27, 2012 4:37:24 GMT
While I understand the original tread was locked as it suffiently answered the questions, I still thought it was a helpful resource showing how to use the program. I hope its restored (locked or unlocked) so people could use it as a resource.
|
|
|
Post by Giles Sparrow on Feb 27, 2012 7:46:13 GMT
In a recent thread somebody kindly posted details of how to use PhotoShop to export an image in the .Y format needed by the Colour Recovery program. Finding myself in need of this information I've been searching for the thread, but I can't find it anywhere! Can somebody help me locate the thread, or failing that would the person who posted the information be so kind as to do so again? Richard. Guilty, m'lud! As others have pointed out that thread seems to have been disappeared,and I can't even find my original in my own posting history. What's more, the basics seems to have been covered by the thread already. *However*, here's a brief recap from memory, just so it's on the record somewhere that hopefully remains accessible. 0) Depending on the source of your image (e.g. DVD screen grab or scan, always avoid saving it in a format that introduces further compression. If you *must* save as a jpeg, then make sure it's maximum file size/quality and minimum compression. Ideally save to an uncompressed .tif or some other lossless file format. 1) Open your image into Photoshop. If you use your computer's "print/copy screen" function to do a screen grab, once you've copied the image you can make a new file in Photoshop (Menu: File>New) and it should come up with a dialog box that has the correct dimensions for your grab already filled in. You can then just paste the image into the file. 2) Drop out any rogue colour information, since this is likely to be mpeg compression artefacts - Menu:Image>Mode>Grayscale. While you have the Mode menu visible, check that "8 bits/channel" is ticked. 3) Enlarge the image if necessary. The Colour Recovery program takes files with dimensions of 1920x1080 pixels, so use Menu:Image>Image Size. Type the Width (1920) and Height (1080) into place, unchecking the "Constrain Proportions" box if necessary. Make sure the "Resample Image" checkbox is on, and choose an appropriate resampling algorithm from the dropdown. "Nearest neighbour" is likely to be the best method of resampling since it would avoid unnecessary blurring of the chroma dots, but this would be worth experimenting with... 4) Save the image. Menu: File > Save As... Select "Photoshop RAW" from the dropdown list, and crucially, replace the ".raw" file extension in the filename with ".y". Save, and you should have a file that will open in Richard's CR program. Whether there's any chroma dot info in there, of course, is another matter... Cheers, Giles
|
|
|
Post by Richard Russell on Feb 27, 2012 10:14:59 GMT
here's a brief recap from memory, just so it's on the record somewhere that hopefully remains accessible. Thank you so much, that's just what I wanted. It's got that extra detail that can make all the difference. With your permission I'll put a précis on my Colour Recovery Wiki. I don't know who deleted the original thread but I consider it heavy-handed. I can see the reasons for locking it, but it contained useful information (not least your Photoshop instructions) which deserved to be kept available. Richard.
|
|
|
Post by Alex Weidmann on Feb 27, 2012 10:18:09 GMT
Thanks Richard. I'll try those three resamplers and increase the number of taps. Haven't spent any time on optimising the filters yet, as I was much more interested in the field separation effect.
If you have time would welcome your opinion on whether this process would be useful. The details are on the CRWG Wiki current work section, under the title "Extrapolating A More Complete Field Separation". (N.B. The processing scripts can be tested by anyone, as I've published them in full on the Wiki. Once AviSynth is installed you can open the .avs scripts through most media players, or frame-servers like Virtual Dub and HcEnc. They're quite fast too, up to 10fps.)
|
|
|
Post by Richard Russell on Feb 27, 2012 12:38:50 GMT
If you have time would welcome your opinion on whether this process would be useful. The details are on the CRWG Wiki current work section, under the title "Extrapolating A More Complete Field Separation". What is the intended purpose of the 'field separation'? Are you hoping it could replace or supplement the existing VidFIRE process to recover the original 50 Hz motion? I'm not aware that inadequacies of VidFIRE have been identified as a 'problem'. As I think you know, the Film Recorders used 'spot wobble' to reduce the visibility of the line structure, so the two fields are inextricably mixed together on the film anyway. The machines seem to have been quite variable in this respect (the TOTP sequence you are using having untypically little crosstalk) so the opportunities for field separation must be limited. Incidentally, in reaction to your comment at the very bottom of the CRWG page, 288 c/aph (the highest vertical frequency present in a 625-line TV frame) is "well below the Nyquist Limit of an HD 1080-line scan" too! Don't fall into the trap of thinking that the vertical spectrum extends above this frequency. Richard.
|
|
|
Post by Rob Moss on Feb 27, 2012 14:03:46 GMT
I assumed that wanting to be able to unpick the fields was in an attempt to fix out of phase film inserts.
|
|
|
Post by Richard Russell on Feb 27, 2012 14:37:08 GMT
I assumed that wanting to be able to unpick the fields was in an attempt to fix out of phase film inserts. If that was the principal objective, I would have expected Alex to be using just such an out-of-phase film insert as his source material, rather than a studio video shot. Nevertheless it would be interesting to try it, even though the degree of field separation is necessarily limited. Richard.
|
|
|
Post by pelham cort on Feb 27, 2012 14:55:11 GMT
It was me who made the thread and giles gave the info but the thread was locked for some reason. Richard Russell: Can you please tell me where i can find a telerecording of dw episodes with chroma dots on IE: Mind Of Evil Ep 2?? Or could you upload some footage?? Thanks Simon, the thread was presumably locked because it became clogged up with your hand-coloured screen grabs from Timeslip. dude leave my friend alone.he was trying to make it look like the original videotape and you rejected it. you probably tweaked your colour recoveries like on dads army and ep1 of invasion of the dinosaurs.
|
|
|
Post by Alex Weidmann on Feb 27, 2012 15:11:31 GMT
What is the intended purpose of the 'field separation'? Are you hoping it could replace or supplement the existing VidFIRE process to recover the original 50 Hz motion? I'm not aware that inadequacies of VidFIRE have been identified as a 'problem'. Thanks very much for taking a look! The idea was to allow the field-order of out-of-phase film inserts to be repaired, so the double-imaging can be removed or reduced. I imagined this could be useful for bridging short sections of footage where the double-imaging is particularly bad. If I've understood correctly, each frame in a film recording contains a mixture of two frames from the original film insert reel (if it was telecined out of phase). So if the fields in the FR are separated, I reasoned the original film frames from the insert reel could be restored by re-arranging the field-order. Hope I've got that right? As I think you know, the Film Recorders used 'spot wobble' to reduce the visibility of the line structure, so the two fields are inextricably mixed together on the film anyway. The machines seem to have been quite variable in this respect (the TOTP sequence you are using having untypically little crosstalk) so the opportunities for field separation must be limited. Yes that's very true. I suspect it will work on some other FR's; but definitely not all of them. I have a suspicion 377i material may be easier to separate because the scan-lines are fatter. Incidentally, in reaction to your comment at the very bottom of the CRWG page, 288 c/aph (the highest vertical frequency present in a 625-line TV frame) is "well below the Nyquist Limit of an HD 1080-line scan" too! Don't fall into the trap of thinking that the vertical spectrum extends above this frequency. I did think there'd be some energy at the line frequency if the spot-wobble was low enough to allow for partial scan-line separation. If there was no spot-wobble, I thought the line-structure would produce a signal at 0Hz, 573c/aph (assuming 3 lines are cropped in the film frame). In the vertical direction I pictured idealised scan-lines as square-pulses with a slight separation. Sounds like that was not a valid model.
|
|
|
Post by Alex Weidmann on Feb 27, 2012 15:17:21 GMT
I assumed that wanting to be able to unpick the fields was in an attempt to fix out of phase film inserts. If that was the principal objective, I would have expected Alex to be using just such an out-of-phase film insert as his source material, rather than a studio video shot. I did try it on another source containing an out-of-phase film insert; but got no separation. Other than that, I don't have any further sources to test it on unfortunately.
|
|
|
Post by Giles Sparrow on Feb 27, 2012 15:27:24 GMT
Thank you so much, that's just what I wanted. It's got that extra detail that can make all the difference. With your permission I'll put a précis on my Colour Recovery Wiki. Richard. Sure, no problem - happy to help!
|
|
|
Post by Simon Broad on Feb 27, 2012 15:27:38 GMT
Thanks Pelham I was only showing some of my colour recovery pics of and if you dont want threads to be clogged up why allow people to upload picstures? Regards, Simon
|
|
|
Post by Rob Moss on Feb 27, 2012 15:36:50 GMT
Simon, the thread was presumably locked because it became clogged up with your hand-coloured screen grabs from Timeslip. dude leave my friend alone.he was trying to make it look like the original videotape and you rejected it. you probably tweaked your colour recoveries like on dads army and ep1 of invasion of the dinosaurs. You are completely wrong in that assumption. I've never tried the CR software, because I don't have access to high-res scans and I know my limitations, but regardless of that, this thread isn't the place to be posting hand-coloured DVD frames, especially if you're going to try and palm them off as genuine CR results.
|
|
|
Post by Richard Russell on Feb 27, 2012 15:43:25 GMT
If there was no spot-wobble, I thought the line-structure would produce a signal at 0Hz, 573c/aph (assuming 3 lines are cropped in the film frame). In the vertical direction I pictured idealised scan-lines as square-pulses with a slight separation. The crucial thing to realise is that baseband analogue (625-line) TV, looked at as a 2D signal, is sampled vertically at 576 c/aph. So according to sampling theory it cannot contain any vertical information above a frequency of 288 c/aph (the Nyquist limit). Even if the lines are "square-pulses with a slight separation" there is no added information content. It's a bit like upconverting a sampled signal by inserting zeroes between the samples: the sample rate is increased but no information is added. You only need a sampling rate of a little over 576 c/aph to capture all the vertical information, in principle. Richard.
|
|