Content Aware Scaling on Video Layer

Hello,
I hope this is possible because it works really well on still frames but how would I go about applying content aware scaling to a video layer inside PS? Basically I'm trying to use this feature to fake an aspect ratio change where by I have a subject surrounded by a relatively uniform background and I would like to be able to create a protection mask as an alpha layer and have photoshop apply the content aware scaling only to the background(to fill in the empty parts of the rescaled video) and leave the subject untouched.
What I've done is opened a video in photoshop as a layer, created an alpha channel to represent the subject(subject is white, background is black on the created alpha channel), changed the canvas size to the new aspect ratio which obviously created a tranparent frame around the original video frame. I then tried to apply content aware scaling to the video, it converted the layer to a smart object then displayed the free transform handles. However, when I selected my alpha channel as the protection area, the content aware scaling seemed to ignore it and went on to distort the entire frame. Now I tried this with still frame from the same video with the same mask/alpha channel and it worked as expected just didn't seem to do it with video.
The idea, if it would be possible to do this in PS, is to rotoscope the subject roughly in AE and generate a matte pass which can then be used as the acompanying alpha layer of the video so then the content aware scaling can be applied to each invidividual frame.
Is this possible??

No that's not what I was trying to do, I mask the central subject of my frame and as I said it was surrounded by a fairly uniform background(main subject is a rock formation in the middle of the ocean) I want to change the aspect ratio of that frame, so I created an alpha mask to cover the main subject, the surrounding background was not covered in this alpha channel mask. This way when I scale using the content aware feature all that happens is the background(the water) is streteched evenly with lesser distorion(than normal scaling) while the masked main subject is "avoided" with completely no distrotion. I think the key point you missed in my description is that I am in fact creating an alpha matte for only my subject not the entire frame and using that in the "protect" option of content aware scaling.
The result is essentially a frame aspect ratio change that does not sacrifice the resolution around the main subject(because the main subject remains of the same size and PS doesn't have to interpolate any pixels...this is obviously is not the same as just resizng and cropping where the subject and background are streched evenly to fill an essentially larger area of the screen causing "softness"). The result using this method is also much more predictable from frame to frame than content aware filling because it's taking information from the current frame and using is to fill the empty background resulting from the aspect ratio change. I know this works because I tried it on individual still frames from the video then compiled them back in on AfterEffects...thats not my problem. My problem is when you try to natively try to do this inside PS on a video layer(rather than a still frame) it doesn't work and I'm not sure if this is a bug or if video content aware streching is just not supported.I have CS6 Production Premium.

Similar Messages

How to only have 1 image with content aware scaling

My knowledge (lack of, really) with content aware scaling is that 1st you create a new layer of the background and then rescale accordingly. I did the following: -1- Created a 2nd layer -2 used quick selection tool on my main subject ( a person) and named it. -3- deselected the selection -4- Edit menu --> Content aware scale -5- resized it to something smaller (chose my previous named selection channel). -6- Pressed "Enter"
The result is that I can now see 2 images; the smaller sitting on the larger. What now? All I want is the smaller image. If I throw away the original layer what I end up is the new layer with the white and gray default background showing as part of the image (which ends up being printed as well or if saved looking as is with the grey-white. Is what I should do the obvious--cut out via cropping the un-needed (gray-white) part? Or am I missing a step somewhere?
Thank you

My changing of image size had always been either by using the crop tool (which of course also removes parts of the image), or going to the top menu and choosing 'Iimage Resize' ( Alt + Ctrl + I). So using any method other than the one described alway had 'Content Aware Resize' greyed out and therefore unuseable hence me thinking a 2nd layer needed to be done. I had not been aware to use the marquee tool--so thanks. Either way though, when scaling down, it seems that cropping needs to bee done just the same. I was hoping there was a method where this was not required and could be done directly. Your method was defnitely helpful and saved me a step nevertheless so thanks. BTW, if I may, why does everywhere I go and have seen always show to create an extra layer? Your method seems to be better and actually shows creatin an extra layer to be totally useless so that is why I am asking since it seems to be totally not needed.

Problems with Clone Tool / Content Aware Scaling in CS4 Mac

I've been using PS for about 1 year and a half. Sort of an Advanced beginner. One of the things I love is the clone tool. Since I've upgraded to CS4 I've had nothing but problems with the Clone tool. When I go over an area it doesn't show anything and the either makes the adjustments after a long delay or it goes black and then comes back with the adjustment. Sometimes when I'm dragging the brush the whole picture goes away replaced by a smear of color noise. It really makes the tool unusable. In many cases I've gone back to CS3.
I also have similar problems with the content aware scaling. Never been able to get it to work. Basically some of the nicest new features of CS4 are worthless for me. Tried looking at Adobe solution and searching the forums but I haven't seen any exactly similar problems of solutions.
Any help appreciated. I have a white 24' Intel Mac with a 2.33 GHz Intel Core 2 Dual with 3GB RAM.
Alan

As always the various User Forums are fantastic. Adobe (and other) get everyone to do their Tech Service for them. Thanks for the help.
So, two new pieces of info. When I disable Open GL everything works fine. (Sorry about the TIFF, I just didn't catch it) So in essence this solves my problem.
Also PS CS4 works perfectly on my new Aluminum MacBook. So clearly it is an issue with the video board on my iMac.
This brings me to the wrap-up question. Should PS CS4 theoretically work with Open GL on my iMac or is it a know problem with CS4 and older graphics boards? I'm assuming my board supports Open GL.
This is the Profiler data on the Graphics Card
Chipset Model: NVIDIA GeForce 7600 GT
Type: Display
Bus: PCIe
PCIe Lane Width: x16
VRAM (Total): 256 MB
Vendor: NVIDIA (0x10de)
Device ID: 0x0391
Revision ID: 0x00a1
ROM Revision: 3021
Displays:
iMac:
Resolution: 1920 x 1200
Depth: 32-bit Color
Core Image: Hardware Accelerated
Main Display: Yes
Mirror: Off
Online: Yes
Quartz Extreme: Supported
Built-In: Yes
Some of the Open GL features are nice and wouldn't mind having them but not worth buying a new Mac for. So the questions is: Is the some kind of upgrade / driver I can load to solve this problem.
Again Thanks for the help

CS4: Content Aware scaling never works....

Quad 2.66 Xeon MacPro
16GB RAM
Dedicated 250GB internal Scratch Drive (Not a volume)
2 nVidia 512mb 8800GT video cards.
Whenever I choose "Content Aware Scale" the transform handles appear, I alter the image, ht enter to commit and I get this:
>Could not complete your request because there is not enough memory (RAM).
Huh?
The file I'm working on currently is only 37.5mb in size, 2599x2592 pixels, two layers ...
But this error presents itself with each and every attempt to use Content-Aware scale.
Any help?

so I'm not the only one who has problems with content aware scaling -- but there is no current discussion about it and I'm not sure what to do now.
Im on a mac book pro that i bought october 2008 with 4 Gb RAM, with a 2.8 GHz Intel core 2 duo processor. I run content aware scaling on a file and get the warning dialogue: Could not complete your request because there is not enough memory (RAM).
People have talked about CAScaling not working well with Bigger tiles plug in activated. as far as I can see I do not even have that plug in. I go to apps>PS CS4 >plug-ins > extentions -- and it's not there -- I see plug ins for AltiVecCore, Enable Async IO, FastCore MMXCore, Multi Processor Support. PPCCore, and scripting support -- but not a Bigger tiles plug in which I still have in my previous copy of CS3 -- but not CS4.
So I'm wondering what's up - how to solve this issue and get back to work -- and just how I should have my PS performance preferences set up for best performance. Anyone out there with a clue?
thanks,
owen

Content aware scaling

I am trying to use "content aware scaling" in Photoshop CS4. But every time I try to use it, I get a "program error" messaage. I select the image, try and resize it, but when I click for content aware scaling to work, the error message comes up. ( I am using windows XP, if that means anything).

May I respectfully suggest you polish up your skills on searching for things on the Internet...
Google turned this up in about one second after I entered "picture package CS4 download" into the little box at the upper-right corner of Internet Explorer: http://www.adobe.com/support/downloads/detail.jsp?ftpID=4048
-Noel

Using content aware with a new layer

I am trying to use content aware with a new layer which i have made a smart object but when I select fill it is greyed out. why?.

CAF doesn't have a Sample all layers option, and needs source pixels to work on. The best you can do is select those parts of the background layer that you are happy to be sampled from, and copy them to a new layer and use CAF on that layer. This prevents gaps in the landscape being filled with a foregrounsd subject's head!
You can then add a layer mask to that layer to the new layer to use only those parts you need.

Content Aware Scaling and not enough RAM

I'm on a mac book pro that i bought october 2008 with 4 Gb RAM, with a 2.8 GHz Intel core 2 duo processor. I run content aware scaling on a file make my adjustments, click return and get the warning dialogue: “Could not complete your request because there is not enough memory (RAM).”
In earlier adobe forums people have talked about CAScaling not working well with Bigger tiles plug in activated. as far as I can see I do not even have that plug in. I go to apps>PS CS4 >plug-ins > extentions -- and it's not there -- I see plug ins for AltiVecCore, Enable Async IO, FastCore MMXCore, Multi Processor Support. PPCCore, and scripting support -- but not a Bigger tiles plug in which I still have in my previous copy of CS3 -- but not CS4.
So I'm wondering what's up - how to solve this issue and get back to work -- and just how I should have my PS performance preferences set up for best performance. Anyone out there with a clue?
thanks,
owen

I think I discovered the source of my problem -- doing too many things at once -- I did however , as you suggested, Buko, allot 90% of my RAM to photoshop. My image was a panorama made from seven RAW files shot with my Canon G-10 -- about 15 inches by 63 inches @ 240 ppi. I first stitched them together out of Lightroom -- no problem. But of course the edges were rounded, and I wanted to stretch my sky and foreground to fill the image rectangle without any apparent distortion. I figured CAS was perfect for this... and so it is, once I figured things out.
It is now my intent to stitch the pano every way possible in CS4 and see which one looks best in this particular instance. I may have multiple favorite methodologies, and discover which procedure fits what kind of situation best. My problem, I think, was that in applying the CA scaling transformation, I pulled on all four sides of the image - not just one. But PS and CAS worked just fine if I transformed each edge or side separately. Perhaps it works fine pulling and transforming on all sides if I'm on an 8 core mac pro tower with 16 gigs of RAM -- we'll see -- but for now -- I'll try it separately with each edge @ 100% RAM allowance -- to see if that had anything to do with my problem -- otherwise I'll keep it in the optimum range, and transform one side at a time.
For me this file was a test -- I have a panorama of an image looking over the rooftops of Siena, Italy to print next, made from 17 frames from my 5D mark 2 and that file is 3.3 Gb and took my little laptop over an hour to render as a pano -- but it did-- ( thank you apple -- thank you Intel, thank you Mr Knoll & co...) and will need some stretching to fill the frame -- one side at a time - I suspect.
So merci beaucoup, Buko!
... and thanks Mr Schewe for the the tip, doode -- you know what I'm talking about. ;^) and thanks to the engineers and geniuses at Adobe who make this magic possible.
peace,
oc

Content aware scaling - getting very poor results for architecture

I am evaluating CS4 to see if I need to upgrade. I am not excited by the UI changes form CS3 so that is not a reason alone to upgrade for me. I am hoping that content aware scaling is that reason. I shoot architecture and many times there is extreme edge distortion from using really wide angle lenses. I try not to use those lenses but sometimes it is unavoidable. I am not talking about barrel or pincushion distortion. Alot of times there is a chair or a table on the edge of frame that gets stretched and it takes alot of scaling and warping to make it look not as bad. I tried to fix it with content aware scaling using a protection channel but the results are the same as using regular scaling. Am I doing something wrong or expecting too much? Any help would be appreciated.

if quick selection can select an object like a chair or table without too much difficulty, I thought content aware scaling would recognize those elements too like it supposedly does people. if it can only work with vague content like skies, ocean etc than I have no use for it.
and if thats the case i would skip this update and wait for CS5, i don't see much else to get me to update. unless there were other transform improvements.

Content Aware Scaling--Upscaling isn't working?

I am using content aware scaling. Everything works great until I try to upscale the image by dragging the corners. When I try to transform to the larger size, the part that has been widened is "hidden." I know that it is larger because I can "move" it within the white box that marks the original size, but I can't make the complete upscaled image appear. Can anyone help with this? Thanks!

After content aware scale, go to Image>Reveal All
and see if that helps.
MTSTUNER

Content Aware Scale shortcut duplicates Layer ?

There seems to be a difference between accessing Content Aware Scale via the key shortcut and the Edit menu using CS5:
- If I use the Edit menu to call up the Content Aware Scale command on a Layer or background it edits the item directly, replacing the original image.
- If I use the shortcut Cmd + Alt + Shift +C to access the Content Aware Scale command, it gives me a duplicate Layer, or duplicates the image as I carry out a transform instead.
Is this a feature by design ?
thanks
Photoshop CS5 Extended 12.0.1 x64
Mac Pro 2.8ghz quad, 8gb ram, 10.5.8

Anyone ?
If it's not a feature it's annoying and should be fixed.

Can't execute content aware scaling - out of RAM message

Cs4 extended. "bigtiles" plug in not loaded. This is trial version, I haven't bought upgrade yet......but this feature is key for me.
Vista 32 bit, 4gb ram, 32gb available on HD
I keep getting the "can't complete because not enough RAM" message
I tried a)converting to 8 bit file. working file size is 41mb
b)increasing amount of ram used by PS to 80%
c)removing most other processes from running.
efficiency still show up at 100% during process.
Anyone have any ideas? shouldnt my system be able to handle this?
thank you

I don't think the scaling is meant for alphalized/transparent content and selections. After all, it is based on detecting contiguous areas and unique structures in the visible pixels... You may simply need to adapt your workflow and way of thinking for using this feature most efficiently.
Mylenium

Content Aware on Panorama

I was very excited to see the content-aware sneak peek video, where Photoshop automatically filled in the edges of the panorama photo by just selecting the empty space. I was very anxious to give it a go, but was disappointed with the results, or lack thereof.
I stitched together 7 photos that I took and merged the layers. I had to merge the layers because it would not allow me to use the magic wand tool when there were 7 linked layers. When I select the empty space and press delete, nothing happens.
I have noticed if I "Layer by Copy" on other pictures, content-aware doesn't come up as an option on the copied layer. So, I'm guessing my problem with the panoramic picture is that I merged the layers?
Does anyone have ay suggestions as to what I'm doing wrong? He makes it look soo easy in the video.
Thanks!

Layer-Flatten to "merge the layers", select the edges, then use the Delete key. The key here is for the image to be the Background layer. Upon pressing the Delete key it will pop up a menu asking you how you want to fill the parts you have selected, with Content Aware being the default.
You can also bring up the fill dialog when deleting on a bona fide layer by pressing Shift + Backspace (i.e., the Delete key at the upper-right of the main part of your keyboard).
-Noel

CS5 Content Aware Fill

Demos of the new Photoshop CS5 "content aware fill" feature on the Youtubes -- it looks impressive.
I know " Past performance is not necessarily indicative of future results", but any guesses on when (or if) this feature would hit the PS Elements code stream?

This is looking like an advanced "must-have" type of feature so my guess is Adobe will keep it at the premium Photoshop level at least until CS6.
Note that Content Aware Scaling was introduced in Photoshop CS4, which came out at around the same time as Photoshop Elements 7 in the autumn of 2008 - but this other related feature wasn't added to Photoshop Elements 8 last fall. Just a possible analogy.

Remove a logo from a video clip with content aware fill?

I know you can remove logo in photos using content aware fill, but you can't do it to a video clip. You can only do it to a frame, but you can't apply it to the entire video. I know photoshop is not a video editing software, but having this sort of feature makes sense for photoshop users especially with content aware fill.

I'm a career software architect engineer of 35 years, and I have to side with Howard on this one.
Like Howard, I feel you're thinking about this way too simplistically. Trying to work out what best to fill a part of an image with in 3 dimensions (x, y, and time) and have it look believable almost boggles the mind, especially thinking of a general case (your logo example is far to simplistic in itself). The task is far from a simple looping repeat of what is being done 2 dimensionally on a still frame. Frankly, it's more likely related to what's done with CGI, where scenes are modeled in 3D (4D actually) so certain elements can be replaced believably.
Think about it... If CAF is presently replacing part of an image with something it finds through some heuristics in another part of the image, and the stuff in that part of each frame is changing - and it would, because this is video, not still imagery - then it will select differently every time. What makes you think it would select something that even resembles what it selected for another frame? What you'd end up with is a frame by frame mishmash that would just make the outline of the selection obvious.
Regarding CAF applied to individual frames, then your finding the result isn't coming out as expected in all the frames of the video (e.g., when you're testing this concept) may indicate a problem with the way you're using the tool, or possibly even a bug. You might want to describe step by step what you're doing and show the result you're [not] getting.
-Noel

How Content-Aware Fill can help with Video | Short and Suite | Adobe TV

Content-Aware fill in Photoshop is a powerful tool for generating clean plates in Premiere Pro. Karl Soule’ provides some examples.
http://adobe.ly/zpuKyy

Great idea that I can definitely apply to some video that has a similar (set of!) problems. Thanks Karl!

Content Aware Scaling on Video Layer

Similar Messages

Maybe you are looking for