Magic Lantern Forum

In my tests with T3i 600D shooting 768 x 423 @37,7fps and 960 x 544 @ 24fps and upscaling to 1920 x 1080 I found the best solution for removing aliasing:

Red Giant Magic Bullet Instant HD 1.2 (99 usd plug-in)

Best settings for 768x432 and 960x544 resolutions:

output size: custom
1920 x **** (do your aspect ratio calculations) for 960x544 it is 1920x1088, for 768x432 it is 1920x1080
lock aspect ratio: off
filter type: better (important, "better" is even better than "best")
sharpness: 13 (more than this hurts the image)
quality: 8 (more than this hurts the image)
antialiasing: 10 (the maximum possible)

these settings I found after try everything with careful comparisons

you will not believe your eyes.

ACR resizes it even nicer.

Aliasing can't be removed in post without breaking the Nyquist theorem (http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem) ;)

(which is on the same level as breaking physics laws)

@1%, @a1ex,

Hi guys, did you try it? the results are amazing, it deserves a try...

I did a look at the theorem, but my maths is not enough to understand... maybe the Red Giant did a break on it?

Please, can you share how to make the upscaling in ACR to get good results? It will be useful for lots of people...

thanks!

I did another test and realized that apply two instances of the filter makes things even better:

first instant hd filter settings:
custom
1280x720
better
00
00
10

second instant hd settings
custom
1920x**** (height depends on what resolution you are shooting, for 960x544 = 1920 x 1088, for 768x432 = 1920x1080)
better
08
08
10

I did some more tests comparing apply the instant hd once or twice and I realized that twice adds some artifacts to the image, so apply once will be better. apply it once there is no artifacts.

Do you have any screenshots for us to compare?

There is another plug-in called ReSizer which comes togheter in the Instant HD package. It works great also, and delivers slightly different results, and it needs a small amount off sharpen filter also. you can compare both and decide what you prefer... for my taste the instanthd is better than resizer for anti-aliasing results.

here is a crop from a small part of the image for comparison, (240x150 pixels section from the 1920x1080 image). it is not perfect, but it is much better:

Quote from: 1% on June 07, 2013, 06:30:06 AM
ACR resizes it even nicer.

How are you resizing in ACR? I can't:

I had good results with this (http://avisynth.org/mediawiki/AntiAliasing) back in the day.

If you're handy with avisynth, you can try changing the resizing algorithms in the script for better results.

Apply a good sharpener (http://avisynth.org/mediawiki/LSFmod) at the end of your script to break the laws of physics :P

Be prepared to significantly increase your processing time.  But hey, quality comes at a cost ;)

Improving results for 960x544@24p...

All the settings in Instant HD in my previous posts was found for ACR sharpening in AMOUNT=25, RADIUS=1, DETAIL=25

I realized if all sharpening in ACR is set to 0 (zero), Instant HD makes no difference because there is no significant aliasing in the image to correct, but this way, the overall looking is too much soft...

So I started to find the best balance between ACR sharpening and Instant HD correction...

Best results for 960x544 upscaling to 1920x1088 I found was:
setting ACR sharpening AMOUNT=30, RADIUS=1, DETAIL=10
setting Instant HD to sharpening 13, quality 08, anti-aliasing=10, mode=better
more or less the same thing as before, no other sharpen was used
if ACR is the same, this also works in Instant HD: sharpening=00, quality=08, anti-aliasing=06, mode=better

Quote from: apefos on June 07, 2013, 07:10:30 PM
I realized if all sharpening in ACR is set to 0 (zero), Instant HD makes no difference because there is no significant aliasing in the image to correct

I am at an absolute loss to understand why you would increase aliasing just to later find the best way to remove it  :o

Find a better resizing algorithm that doesn't introduce aliasing in the first place ;)

It is so simple:

If you do not do sharpening in the image, it will be too much soft, no good to look at.

But there is a problem: every sharpening filter, effect or plugin from any software will increase aliasing. So this is the challenge: to do some amount of sharpening without hurt the image with aliasing.

Upscaling will magnify the image, so imperfections will be more easy to see. So to do sharpening and upscaling at the same time is something very difficult to do considering we want good results.

In my tests I found the Instant HD upscaling is the best because it corrects the aliasing found in low resolution video.

In my tests I found ACR sharpening is the best results to my eye, maybe because it is applied in the raw file using all the bits before converting.

So, I did tests in three way of doing the sharpening:
1- in ACR before Instant HD upscaling
2- inside Instant HD at the moment it does the upscaling (it has a sharpen slider)
3- using sharpen effect after Instant HD upscaling

The best results was from option 1: apply sharpening in ACR before upscaling and do upscaling in timeline using Instant HD.
Maybe this was the best result because sharpening in ACR uses all the power from raw file and doing upscaling with Instant HD uses the power of it's anti-aliasing which is very good.

As you can see I did try all the options and this was the best results. Using sharpening slider  inside Instant HD is optional, it is a matter of taste because sharpening was already done in ACR, so the different settings in Instant HD in my previous post is for you to choose the best for your eyes.

I just came across this optical filter solution for moire. I'm intrigued:

http://recordingsofnature.wordpress.com/2012/12/04/cyl-filter-moire-and-aliasing-filter-for-hd-dslr-video/

This one is new to me. I've seen the  Mosaic Engineering AA filters, but it's my understanding that they won't work properly at resolutions other than 1080P.

Jason

Apefos.  Could you be so kind as to upload a source image (dng please),  and a image that you have upsized and sharpend?

am i correct in assuming that in 10x mode you don't get any moire or aliasing since there is no line skipping ?

i understand that it is using just a cropped portion of the sensor.

So is it just the footage I have or does ACR process 2003 handle upscaling MUCH better? Someone wanna test it out?

Quote from: Audionut on June 07, 2013, 09:05:29 PM
Apefos.  Could you be so kind as to upload a source image (dng please),  and a image that you have upsized and sharpend?

Would be nice to see this actually.

I already uploaded an image in the post number 7 showing the difference between "Instant HD upscaling" and "default upscaling"

Both images was identically sharpened in ACR before the upscaling. The settings for tweaking these images are described in my posts here in this topic, read them all.

Upscale vs stock test-

http://www.magiclantern.fm/forum/index.php?topic=6268.0

Please 1% and Apefos, see this document in which I compared several algorithms. The ACR and AE use the Bicubic algorithm, technically inferior to all others (except bilinear). The plugin red giant probably uses the Lanczos, Hungarian algorithm.
The best interpolation algorithm until now demonstrably is Smooth Spline. See it all in the document (see at 400%):

https://www.box.com/s/q1i471h34lqm50c1v930

Quote from: apefos on June 07, 2013, 11:35:24 PM
I already uploaded an image in the post number 7 showing the difference between "Instant HD upscaling" and "default upscaling"

Yes, but I would like an untouched DNG so that I can perform my own tests please.
I own a 5D3 that doesn't have aliasing problems.  I guess I could run my tests on my own footage shot at a low resolution, but I envision that someone will come along and tell me that my tests don't count cause it was with a 5D3.

Quote from: jchristman on June 07, 2013, 08:40:32 PM
I just came across this optical filter solution for moire. I'm intrigued:

http://recordingsofnature.wordpress.com/2012/12/04/cyl-filter-moire-and-aliasing-filter-for-hd-dslr-video/

Since there are lines skipped causing the aliasing, the problem is that some light reaches one pixel*, some light reaches the gap in the middle, then some light reaches the next pixel across.
So what an aliasing filter does is blur the light coming to the pixels, so the lost colour information that otherwise gets sent to the 'gap' is included.

Which obviously reduces sharpness and contrast to some extent, but the idea of the aliasing filter is to reduce it by the exact minimum amount required for sensor size and how much line skipping it's doing so it's not noticably blurry...

Could you not acheive exactly the same thing, by shifting the lense focus by a tiny amount forward or backward from a 'perfect' focus and then sharpening the image afterwards?
I know that sounds pretty blasphemous to a photography crowd to intentionally misfocus, but maybe worth experimenting with.
Or just smear some butter on the front of your lense, hahaha.

* = I realise this isnt the correct terminology, not sure what the RGBG cluster type thing is called that recieves light on the sensor.

Infact this is exactly how I shoot computer monitors at work. Often you find a moire effect when the screen pitch matches that of the camera resolution (a form of extreme brick wall). The easiest way to solve this is to back off focus just a little bit. But sure you get a little contamination in the blacks/edges.

Quote from: Audionut on June 07, 2013, 05:40:37 PM
I had good results with this (http://avisynth.org/mediawiki/AntiAliasing) back in the day.

If you're handy with avisynth, you can try changing the resizing algorithms in the script for better results.

Apply a good sharpener (http://avisynth.org/mediawiki/LSFmod) at the end of your script to break the laws of physics :P

Be prepared to significantly increase your processing time.  But hey, quality comes at a cost ;)

I think I fooled around with some AVS AA filters for this kind of video, but I don't recall getting anything satisfactory at the end. I probably didn't use them correctly though. I wonder if an implementation of Nvidia's FXAA could do anything about it. It's a post-AA so it doesn't require real-time rendering to work. I searched for info about it a while ago on Doom9's forum with not much luck.

Quote from: 3pointedit on June 08, 2013, 11:50:15 AM
Infact this is exactly how I shoot computer monitors at work. Often you find a moire effect when the screen pitch matches that of the camera resolution (a form of extreme brick wall). The easiest way to solve this is to back off focus just a little bit. But sure you get a little contamination in the blacks/edges.

Or perhaps, (if lighting allows) stop down as far as you can go, to introduce diffraction... I think that's essentially the same as what the aliasing filter does.

Quote from: Roman on June 08, 2013, 02:11:31 PM
Or perhaps, (if lighting allows) stop down as far as you can go, to introduce diffraction... I think that's essentially the same as what the aliasing filter does.

I tested that on the 50d but it looks very bad as you get everything in focus and you start getting aliasing and halos everywhere. Best thing is to have just a few objects in focus and the rest slightly out of focus, this makes everything nice.

The optical filter have some advices to use in specific lenses and distances, needs tests... Maybe VAF-Txi will be better option, but I do not have it and do not know if it will be useful for 960x544 cropped sensor, needs tests...

I agree when the image starts to go out of focus the aliasing decrease and disappears. In low resolutions raw shooting a very small amount of "out of focus" can help.

I saw the algorithm test. I believe the Instant HD is different... I think it has some proprietary algorithm which corrects the aliasing in the edges and keep the rest of image untouched. In the same way the Power Retouche plugin does. I tested the power retouche, but the render time is infinitely slow and results are not better than Instant HD, (at this moment it is just 32bits, a 64bits version is on it's way), see:

http://powerretouche.com/Antialias_plugin_tutorial.htm

Here it is a 600D dng file, just some wires against the sky for aliasing tests, do your own settings in ACR:

http://www.apefos.com/upload/000240.dng (http://www.apefos.com/upload/000240.dng)

I found a cool one last night, Alien Skin Blow Up. Anyone used it before?

Interesting! Deserves a try!

I found this page also, lots of comparisons in there (it seems these people did a try in almost everything):

http://www.infognition.com/articles/video_upsize_methods_comparison.html

http://www.infognition.com/articles/video_resize_shootout.html

others:

http://www.fredmiranda.com/RP/

http://after-effects.softalizer.com/

Check out this free plugin for AE, uses b-spline.

https://sites.google.com/site/katisvideotoolbox/smooth-bspline-scaler

Alien Skin BlowUp3 turns T3i 600D into a Pocket Cinema Camera!

No aliasing, just NOTHING, NADA!

Upscales PERFECTLY, much better than InstantHD.

Only wires against sky shows a small amount of aliasing (much less than InstantHD), but this is expected.

Worth every penny!

(Just some advices: it is slower than InstantHD, rendering upscaling frames takes long time... and batch processing just works in LightRoom using still images, batch processing does not work in Photoshop, it does not work in video files also. My tests was done in jpg files exported from the dng, I do not know if it can be applied to dng files in LightRoom, needs tests)

I'll have to check it out then, that free one was crap lol. ACR was better.

I take that back, didn't realize OpenGL had to be off. Gonna test it some more.

more options:

http://www.benvista.com/photozoompro

http://www.fixerlabs.com/EN/upsize/sizefixer/sizefixerxl.htm

(the last one is supposed to be better than blowup3, but slower)

Here two screen grabs, before/after defringe in ACR to get rid of false color detail:
without [50d at 1596x636]
http://farm9.staticflickr.com/8404/8994151471_89d1925783_o.jpg
with
http://farm8.staticflickr.com/7400/8994152585_2664ed408c_o.jpg

wow, excelente results with defringe!

a review comparing blowup3 and sizefixer, maybe the new version of sizefixer is faster, the article is old...

http://photo.it-enquirer.com/2011/11/review-alien-skin-blow-up-3-vs-sizefixer-xl/

a comparison between sizefixer and blowup3. blowup3 is more plastic look, but sizefixer does not remove aliasing completely. add a small amount of film grain simulation can be a good solution for both, to hide the imperfections and bring a more analog film look. both are 960x544 upscaled to 1920x1088.

and here is the instanthd vs default 200%

Blow Up with film grain would look great imo, too bad its not fot AE :(

yes, I already did a try, blowup3 with film grain is perfect! you cannot tell it was shoot in 960x544, looks like fullhd for sure. I think it will be even better if using the defringe in acr before upscaling. I used sharpening 30 / 1 / 10 in acr and worked fine, no need sharpening in blowup3.

sizefixer is a little bit better than instanthd, but just a little bit...

but instanthd works in after, so it is faster and easier to do.

maybe we could do a petition for Alien Skin to develop a version for after...

I found that importing the dng's into photoshop and in the Raw dialog box that opens at the bottom you can change the size, i found that when filming in 1280x720 and changing it in this box and then using an automate script to then descale down to 1920x1080 from say 2034 when upscaled in the raw settings box, it gave really nice clean results at 1080p ill try and post some results.

Please upload some screen shoots showing us how to do  it.  Also tell us the steps to find this scaling dialog box.

Sounds daft but how do i upload an image? sorry new to this forum.

Sure, if you like the results im more than happy to show you how, the samples also have no sharpening added, i left it as standard so you get the idea of what you get.

Ill make a screencapture video, be easier  :D

ok, it will be nice to see the screen capture video!

Almost done mate, uploading now  ;D

Its in process now so heres the link when its ready

http://www.youtube.com/watch?v=Glv8npoNh4c

Hi, thanks for the video, I saw it.

When I open the 960x544 dng in ACR, in the same dialog box you have the resolutions, it does not have other resolution to choose. Maybe I am using an old version?

No problem,
Sorry its not a great tutorial, was just something quick for you.

I think you could have an old version, I'm running 7.4 or it could be that the resolution you are filming in cant be upscaled in ACR, i have some photos that wouldn't for some reason but again that was on an old version.

Quote from: apefos on June 09, 2013, 04:33:56 PM
Hi, thanks for the video, I saw it.

When I open the 960x544 dng in ACR, in the same dialog box you have the resolutions, it does not have other resolution to choose. Maybe I am using an old version?

The same happened to me. This resolution doesn't allow upscaling. Try 1024x480. It will allow you to upscale to 2048x904.

Thanks!

I can record 1024x512@24fps no problem...

The upscaling from 1024x512 to 2048x1024 in ACR works, but it looks bad... lots of aliasing, the other upscaling options in this topic looks better.

Hello guys!
So, sorry for not answering until now.
What happens is that the people here in the forum is trying to find a software that is better but not studying what happens when using an interpolation algorithm.

If we read about algorithms, we see that there are different way to work with the image. For example: bicubic analyzes four pixels around and creates a new pixel around these.
What I meant by the test that asked them to look at was that, although there are many excellent software, today none can compete with the algorithm "Smooth-Spline" that is scientifically superior. He analyzes the pixels and creates 256 others, ie, retains all the details.

What anti-aliasing filters are doing is only give smooth the edges and then putting sharpen with a large radius, giving the feeling of being less aliasing, which is actually a lie. Maybe there are algorithms that read the edges and make a selective interpolation of these pixels.

Although I am not an expert, I would say the best is still the variation of S-Spline created by the PhotoZoom, the S-Spline Max.
See the software here: http://www.benvista.com/photozoompro

I'll do a comparison with all the other time software mentioned in this discussion.

I did a test now with the PhotoZoomPro 5 and yes, it is great.

the S-Spline Max Algorithm delivers great results in generic mode.

It deserves a careful comparison with SizeFixer and BlowUp3

Waiting for your tests...

This is the result from PhotoZoom Pro 5 with S-Spline Max in generic settings:

To my eyes it is not as powerful as blowup3 in anti-aliasing, but it is less plastic look, maybe a good balance:

Yes, Blowup3 provides a result with a more "beautiful", to give the effect that I mentioned in the last answer. But, note carefully the right image in the lower right corner. In PhotoZoom can see the leaves entwining, now BlowUp3 information becomes confused.

I maintain my opinion that the best result we achieve is using S-Spline Max in combination with other interpolation algorithm for selective edges, for the purpose of anti-aliasing (the effect would be temporal also).

Does upscaling actually make a difference in picture quality? For example which would look sharper: upscaled 1792x672 to 1920x whatever or just the original 1792x672. This is viewed on a 1080p screen...

Please, see the comparison "PhotoZoom PRO 5 vs. BlowUp 3", here (download to see on PS): https://www.box.com/s/1f5xiflhlw4jxqueh5z5 (https://www.box.com/s/1f5xiflhlw4jxqueh5z5)

Note that the S-Spline (PhotoZoom) offers better quality, keeping all information.
Now that we know the best algorithm to resize, the next step would be to find plugins for anti-aliasing. As rightly suggested Apefos the software Power Retouche can offer good results ... I could not test it because I'm 64bit and I have the 32bit version of Photoshop.

Anyone willing to test?

I saw the psd test, for my eyes both looks pretty much the same... only difference is one has more sharpening than other, if the amount of sharpening was the same, they would look almost identical... (I see a difference in the post shape also)

I am doing a test in a jpg sequence to render a video and see blowup3 and photozoom in motion to perceive it better.

As you can read in one of my early posts, I already tested the Power Retouche. It is very very slow in render time... almost impossible to wait... It is not improved to use GPU or multicore CPU. What it does is to find the edges where there are aliasing and smooths the edges without hurt the overall image. But the results are not so much different from PhotoZomm, maybe even worse. I believe if we use photozoom or blowup3 or even instanthd, the Retouche does not make sense... but maybe you can do some tests and try to find some settings which worth using it, apply retouche after photozoom to see if it makes difference...

I am realizing the amount of time and work to shoot raw video considering upscaling and sound sync can be a nightmare... maybe good picture styles in h264 fullhd with some sharpening and grading can be still usefull for easy and fast work.

So ... I think we should organize it here.
What we are discussing with such software is basically resize and interpolation algorithms. The anti-aliasing filters should be applied after resize and not before or during the same.
As you very well emphasized, PhotoZoom has greater clarity, but not clearly false, but clearly "true" because it kept the detalher and that's what we want at this stage.
The second step, now that we found the best resizer, is to find an appropriate software to give smooth the edges. Since you already tested the software already quoted and analyzed the delay is a factor complex to handle, discard the effective use of it. I found other software that I think is appropriate, "PSOFT anti-aliasing" (see here: http://www.psoft.co.jp/en/download/anti-aliasing/ps.html).

I did a video rendering comparison to see the motion using a 960x544@24p RAW video from 600D.

There is no doubt for me anymore. Alien Skin BlowUp3 is the winner. No other can do a upscaling with aliasing free results as it does. also the objects edges does not tremble like in the others. the human face skin texture is good. The small amount of plastic look can be easily solved with a small amount of film grain simulation. It looks like it was shoot with a fullhd camera.

The tips are: enable defringe in ACR to remove color fringing, do the sharpening in ACR (30, 01, 10), do noise reduction in ACR: luma between 30 and 60, chroma 100 (detail zero in both), do the color and luma corrections as you like, render a 960x540 jpg sequence, import all jpg into LightRoom, and apply BlowUp3 using the batch processing (remove all sharpening from it, just enable 200%, or the upscaling amount you need), then render the upscaled jpg sequence into a video file with film grain applied to it.

BlowUp3 render times are faster also: 2,6x faster than PhotoZoom5. PhotoZoom, InstantHD, SizeFixer, PSoft and Retouche do not remove all the aliasing, and the residual aliasing from them is big enough to annoy the eye and small enough to be removed by another filter. For a perfect film look removing all the aliasing is a must and BlowUp3 can do it. Of course in some thin lines and distant trees BlowUp3 also leaves a small amount of aliasing, but I found these situations are impossible to solve. Also BlowUp3 residual aliasing in distant trees and thin lines is always smaller than the other solutions, and the film grain disguises it in a good way, giving an analog chemical film look to the image.

Also remember you can shot 1344x768 @12fps and 1664x944 @8fps in situations where there is not too much motion and convert to 24p with avisynth or twixtor. This just works good for situations with very small motion. 960x544@24p is the way to go for common situations using the tips posted here.

T3i 600D is a low budget "Magic Cinema Camera" now. The worrkflow is a little hard to do, separate audio recording, sync audio in post, long rendering times for dng conversion and upscaling, and so on... Due to this I still believe in good picture styles for fast and easy work in fullhd with sharpening and grading in post. Good to have all the options.

Ok, continue fooling themselves. If you really understand the process that would see the tutorial that just went completely flawed.
Seriously, sharpen before do resize? Come on!  :P

Using a 60D, this is Instant HD on a test image with various resolutions: https://www.youtube.com/watch_popup?v=N54zdFAzGro&vq=hd1080

Other than 5x mode (5 seconds only), aliasing is not acceptable to me.  The 5DMk3 performed better than 60D, but still unacceptable to me.  My cheapo Gopro Black at 2.7k 24fps destroys the 5DMk3 in resolution and aliasing. hehe!

Quote from: apefos on June 10, 2013, 03:25:25 AM
I did a video rendering comparison to see the motion using a 960x544@24p RAW video from 600D.

There is no doubt for me anymore. Alien Skin BlowUp3 is the winner. No other can do a upscaling with aliasing free results as it does. also the objects edges does not tremble like in the others. the human face skin texture is good. The small amount of plastic look can be easily solved with a small amount of film grain simulation. It looks like it was shoot with a fullhd camera.

The tips are: enable defringe in ACR to remove color fringing, do the sharpening in ACR (30, 01, 10), do noise reduction in ACR: luma between 30 and 60, chroma 100 (detail zero in both), do the color and luma corrections as you like, render a 960x540 jpg sequence, import all jpg into LightRoom, and apply BlowUp3 using the batch processing (remove all sharpening from it, just enable 200%, or the upscaling amount you need), then render the upscaled jpg sequence into a video file with film grain applied to it.

BlowUp3 render times are faster also: 2,6x faster than PhotoZoom5. PhotoZoom, InstantHD, SizeFixer, PSoft and Retouche do not remove all the aliasing, and the residual aliasing from them is big enough to annoy the eye and small enough to be removed by another filter. For a perfect film look removing all the aliasing is a must and BlowUp3 can do it. Of course in some thin lines and distant trees BlowUp3 also leaves a small amount of aliasing, but I found these situations are impossible to solve. Also BlowUp3 residual aliasing in distant trees and thin lines is always smaller than the other solutions, and the film grain disguises it in a good way, giving an analog chemical film look to the image.

Also remember you can shot 1344x768 @12fps and 1664x944 @8fps in situations where there is not too much motion and convert to 24p with avisynth or twixtor. This just works good for situations with very small motion. 960x544@24p is the way to go for common situations using the tips posted here.

T3i 600D is a low budget "Magic Cinema Camera" now. The worrkflow is a little hard to do, separate audio recording, sync audio in post, long rendering times for dng conversion and upscaling, and so on... Due to this I still believe in good picture styles for fast and easy work in fullhd with sharpening and grading in post. Good to have all the options.

Whats you're workflow for blowup? I can't seem to find a way to easily batch process all these images. I don't want to import to lightroom.

Quote from: xNiNELiVES on June 10, 2013, 07:17:11 AM
Whats you're workflow for blowup? I can't seem to find a way to easily batch process all these images. I don't want to import to lightroom.

Same things here. Blowup plugin refuse to open "smart objects" :/

Quote from: togg on June 12, 2013, 01:31:54 AM
Same things here. Blowup plugin refuse to open "smart objects" :/

If you want to make this workflow (I do not recommend), export the DNG in Lightroom as TIF or Jpeg ... BlowUp not support smart object.

Quote from: Luiz Roberto dos Santos on June 10, 2013, 04:10:22 AM
Ok, continue fooling themselves. If you really understand the process that would see the tutorial that just went completely flawed.
Seriously, sharpen before do resize? Come on!  :P

I would like to say that I do NOT fool people.

The solution I found was tested and compared with all other options discussed here in this topic. I downloaded a trial and did a try in EVERY software discussed here. Even testing sharpening BEFORE and AFTER upscaling. The comparisons were made ​​in separate dng files and also in dng sequences rendering videos.

Sometimes it is hard for people to accept that someone else found a better solution and people starts getting jealous and speak evil without even testing before speaking. Maybe it is some king of unfair competition... when someone starts to say sh*t about other people's work in an attempt to undermine the reliability and credibility.

Then I recommend to each one to test all methods and software mentioned here to realize that the workflow I found is actually the best.

Of course as I said, it has a little plastic look, but with a small amount of film grain there is no better solution for aliasing removal in upscaling 960x544 to 1920x1088 and crop to 1920x1080, and the final results looks great, better than all other options, without aliasing, without edge trembling, without artifacts.

you need to do the tweaking in ACR as I said in previous post and export the dng sequence as jpg or tif files. you need to create a catalog in lightroom, and import jpg or tif files to it because blowup3 do not works in dng files. then select all and edit in blowup3 and do the batch processing with 200% upscaling.

The sharpening in ACR must be 30 / 01 / 10 / 00 or lower, more than this will hurt the image, see previous posts for noise reduction settings.

Sorry, modesty away for a while, but I am really good in what I do, and I do things in a very scientific way. thanks to all the peolpe who helped here in this topic, and I recomend people who starts to go on in the wrong behavior to rethink things.

I have some hard work to do now, so I will keep away for a while.

Moire is the patterns you see when focussing on say a loudspeaker grill, roof tiles etc.

The best way to disguise Moire is to break up the pattern

The best way to break up a pattern is to sprinkle random noise on it (dithering)

just sayin'

I agree with apefos. We as filmmakers strive to get the best quality from what we have. No reason to insult a man's preferences, this forum is here to help people and give us a place to communicate. If you disagree with someone, provide something to back up your opinions.

-

Quote from: tactzer0 on June 14, 2013, 10:55:13 PM
Description:  550D recording in crop mode (no aliasing from line skipping) @ 1024x448. All images processed through ACR/Photoshop. No noise reduction, no sharpening.  Applied lens correction and defringed all lines. Brought all frames into PhotoZoom and upscaled with S-Spline Max to 2K. Brought those frames into After Effects and used OLMSmoother to reduce aliasing from upscaling to 2K. Exported @ 50 mbps. 18-55mm kit lens zoomed all the way out.

That's really nice when you film your dog from time to time.... but when you have substantial amount of material, you might find yourself spending weeks just to PREPARE the clips to edit...

Thanks anyway, friend !  :)

Quote from: apefos on June 13, 2013, 10:44:10 AM
I am really good in what I do, and I do things in a very scientific way.

0K, I stopped the discussion.

1 - If you really had scientific knowledge would not be giving sharpen before the resize. Apply sharpen before resize is a complete contrast, since you will be increasing the aliasing that is precisely what you want to withdraw;

2 - If you had scientific knowledge, would be speaking in demosaicing and not simple interpolation with Lanczos (BlowUp 3);

3 - Now directing me to the "N/A", I gave you a method that is clearly more applicable, just test it. The goal of interpolation is to conserve as much detail, and not take aliasing, you are confusing things.


I apologize for that, we are here to develop a methodology and not to "fight", but begins from the moment you say you have scientific knowledge, when in fact it is only random testing software, I'm sorry, but I think it is foolish and intolerable.

Test software based on "dcraw" if they want something to pay for something.

The scientific way of obtaining the results you want in image processing is to change the parameters of your workflow in a systematic manner untill you achieve the best looking result. This is of course subjective, but isn't that what matters in image processing?

And it is great to share your results, so that other people can critique and improve on your workflow. What is not scientific, is to state what is "known" without showing anything to back it up. I have read enough scientific literature to know that statements such as "it is widely known"  is often found in poorly written papers.

Quote from: Kim.dh on June 15, 2013, 08:34:44 PM
The scientific way of obtaining the results you want in image processing is to change the parameters of your workflow in a systematic manner untill you achieve the best looking result. This is of course subjective, but isn't that what matters in image processing?

And it is great to share your results, so that other people can critique and improve on your workflow. What is not scientific, is to state what is "known" without showing anything to back it up. I have read enough scientific literature to know that statements such as "it is widely known"  is often found in poorly written papers.

I understand your side. I'm sorry if I was rude, but I think what we're looking for here is not a result "subjectively beautiful", but something that can be proven, technically, be better, and the best is invariable and it is not subjective.
I think this is the purpose of a forum that works with programming and technical data as well-aimed. If not, please, let me know that, I walk away from it.

I found a review comparing four methods discussed here in this topic: Photoshop, Blowup3, Photozoom, Perfectresize.

It will be good for some real examples:

The review: http://www.tipsquirrel.com/resize-software-a-retrospective/ (http://www.tipsquirrel.com/resize-software-a-retrospective/)

Image comparisons (click in the image to zoom in or right click and save as to download):

200%: http://www.tipsquirrel.com/wp-content/uploads/2012/07/200.jpg (http://www.tipsquirrel.com/wp-content/uploads/2012/07/200.jpg)

400%: http://www.tipsquirrel.com/wp-content/uploads/2012/07/400.jpg (http://www.tipsquirrel.com/wp-content/uploads/2012/07/400.jpg)

600%: http://www.tipsquirrel.com/wp-content/uploads/2012/07/600.jpg (http://www.tipsquirrel.com/wp-content/uploads/2012/07/600.jpg)

I tested some methods. To do so, I basically downscaled an image to 50%, upscaled it again with different plugins, and compared the results to the original image. Here's a detail of the comparison.

https://dl.dropboxusercontent.com/u/15481036/upscale.jpg

And here you can download the tiffs: https://dl.dropboxusercontent.com/u/15481036/images%20to%20compare.zip

Blow Up 3 is the only one that does something to aliasing. It does so by "destroying" the image and "creating a new similar one", in my opinion. I don't like the plastic look it gives, and I feel like it is inventing too much information and applying sharpness where there shouldn't be any. However, I wouldn't mind using it for some shots with bad aliasing and then mask out the non-aliased area. To my eye, the winner is S-Spline from Photo Zoom, the best results in both sharp and blurry areas, while respecting the look of the image.

I wrote some code to compare the images in a measurable way. Basically, the code tells you either how many pixels in the original image match the ones in the processed ones, or how different the different pixels are. The results were really disappointing. Basically it looks like only the Nearest Neighbour image shares around 30% of the information with the original image. The rest of them are mostly "invented". This might also be because there seems to be a small offset (less than a pixel?) between the original and the processed images.

Does anyone now of a method to make this comparison useful?

Just updating with new comparison make by Infognition.
Of course the test is questionable, and they should have used SSIM method for analysis, but I think valid post (really slow process, but good result with super-resolution):

http://www.infognition.com/articles/video_resize_shootout.html

For upscaling, one of the eedi (http://forum.doom9.org/showthread.php?t=147695) implementations.

http://avisynth.nl/index.php/AntiAliasing

Quote from: Audionut on June 26, 2014, 01:50:08 AM
For upscaling, one of the eedi (http://forum.doom9.org/showthread.php?t=147695) implementations.

http://avisynth.nl/index.php/AntiAliasing

Try to see the link above... this compare EEDI3 with other methods. As I say, the Super Resolution generate more PSNR.
Of course, the EEDI is freeware (or open source?) script, and Video Enhancer not.

Couple of things about that shootout.

They only measured PSNR.  There are plenty of other metrics (http://www.compression.ru/video/quality_measure/index_en.html) available for testing.  Including some other metrics would have given a greater picture of performance.  PSNR is useful, don't get me wrong, but it is far from the be all, and end all, of video accuracy.

4x resize?  That's a little far removed what I would consider most people are aiming for.  And resizing all the way down to 480x270, has destroyed most detail.  This will favor any resizers which do a half assed job of retaining detail.  Combined with only PSNR measurement   :(

tritical's avisynth plugins are open sourced, the source code is available in downloads from his site (http://bengal.missouri.edu/~kes25c/).