Procedure for blurring faces (VS10 and MSP8)
Procedure for blurring faces (VS10 and MSP8)
Many people ask about this (including myself some time ago.) I *think I have read all the posts about this but am not entirely sure because I don't have the best luck with this forum's search engine.
Anyhow, this procedure seems to work so well that I wonder why/if this hasn't been posted before. There are some similarites to other posts but I haven't seen it this simple.
I d/l and tried the Moving Parts Plugin and wasn't that impressed with it-especially for $100+.
A few people have recommended using the star filter and that works really well. The only problem with the star filter is that you can't really blur or pixelate the area allowing you to see it partially. This way is nearly as simple but with more freedom.
Here's my procedure:
*This works in VS10 and MSP8 [and probably other versions that I haven't tried]
***edit: After much trial and error, I recommend using this method only for VS. The methods listed below work much better if you have MSP.
You can of course blur anything but lets assume we're blurring a face for my example.
1)Open the video of the person moving around.
2)Insert an overlay of the exact video.
3)Apply the crop filter to the overlay vid (or vice versa, it's slightly different for VS10 and MSP8)
4)Adjust the settings of the crop filter so that it only shows the area you want to blur. Insert key frames and make the crop follow the face in motion and size. [It's not as difficult as people make it sound to do this. Create your crop size and position at your first key frame. Move to the end of the significant movement-say the person moves 10 feet to the left. Insert a key frame at that point, crop size and position. At that point the crop/follow work is nearly done for you because the gradual movement and sizing of the key frame1 to key frame 2 will nearly match what the person's face does. You only then need to insert key frames in between keyframe 1 and 2 where it needs tweaking.]
5)Make the cropped out color a bright green or blue. Then chromakey (greenscreen/bluescreen) that color out.
6)Now apply a filter to your cropped video that blurs the person's face. The beauty of it is you can choose, pixelate, blur, ripple etc. ,etc. and adjust them however light or intense you want.
You are then finished. It is about the same amount of work as the star filter procedure but with more "professional" looking results. This is about as easy and good as the moving parts plugin without the $100+ price tag. The only limitation being the fact that your blur box basically has to be some sort of rectangle or square.
Hope this helps...
Anyhow, this procedure seems to work so well that I wonder why/if this hasn't been posted before. There are some similarites to other posts but I haven't seen it this simple.
I d/l and tried the Moving Parts Plugin and wasn't that impressed with it-especially for $100+.
A few people have recommended using the star filter and that works really well. The only problem with the star filter is that you can't really blur or pixelate the area allowing you to see it partially. This way is nearly as simple but with more freedom.
Here's my procedure:
*This works in VS10 and MSP8 [and probably other versions that I haven't tried]
***edit: After much trial and error, I recommend using this method only for VS. The methods listed below work much better if you have MSP.
You can of course blur anything but lets assume we're blurring a face for my example.
1)Open the video of the person moving around.
2)Insert an overlay of the exact video.
3)Apply the crop filter to the overlay vid (or vice versa, it's slightly different for VS10 and MSP8)
4)Adjust the settings of the crop filter so that it only shows the area you want to blur. Insert key frames and make the crop follow the face in motion and size. [It's not as difficult as people make it sound to do this. Create your crop size and position at your first key frame. Move to the end of the significant movement-say the person moves 10 feet to the left. Insert a key frame at that point, crop size and position. At that point the crop/follow work is nearly done for you because the gradual movement and sizing of the key frame1 to key frame 2 will nearly match what the person's face does. You only then need to insert key frames in between keyframe 1 and 2 where it needs tweaking.]
5)Make the cropped out color a bright green or blue. Then chromakey (greenscreen/bluescreen) that color out.
6)Now apply a filter to your cropped video that blurs the person's face. The beauty of it is you can choose, pixelate, blur, ripple etc. ,etc. and adjust them however light or intense you want.
You are then finished. It is about the same amount of work as the star filter procedure but with more "professional" looking results. This is about as easy and good as the moving parts plugin without the $100+ price tag. The only limitation being the fact that your blur box basically has to be some sort of rectangle or square.
Hope this helps...
Last edited by ebone on Tue Mar 20, 2007 12:58 am, edited 1 time in total.
I believe your procedure is similar to what has been recommended before. However, there is a limitation: you are limited to 16 keyframes in the moving path so, if the subject moves a lot within the frame, you may have to divide the clip into 2 or more sub-clips butted on to each other.
My personal preference is to use an ellipse, rather than a cropped rectangle. Draw an ellipse of suitable size on a paper/canvas in your favourite photo editor, a distinctive colour on a transparent background and import it into the timeline (TGA). Drag it to the required positions and size in 2-D MP and set the keyframes. Then add your filters and make to colour transparent.
My personal preference is to use an ellipse, rather than a cropped rectangle. Draw an ellipse of suitable size on a paper/canvas in your favourite photo editor, a distinctive colour on a transparent background and import it into the timeline (TGA). Drag it to the required positions and size in 2-D MP and set the keyframes. Then add your filters and make to colour transparent.
[b][i][color=red]Devil[/color][/i][/b]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
*my reply is screwed up and some of it is in your quote.Devil wrote:I believe your procedure is similar to what has been recommended before.
Yeah, I believe it was somewhat similar to Steve's(?) recommendation but I didn't quite get that one as it involved more steps.
>>>However, there is a limitation: you are limited to 16 keyframes in the moving path so, if the subject moves a lot within the frame, you may have to divide the clip into 2 or more sub-clips butted on to each other.
Yes but that is so simple to do I figured it wasn't even worth mentioning.
>>>My personal preference is to use an ellipse, rather than a cropped rectangle. Draw an ellipse of suitable size on a paper/canvas in your favourite photo editor, a distinctive colour on a transparent background and import it into the timeline (TGA). Drag it to the required positions and size in 2-D MP and set the keyframes. Then add your filters and make to colour transparent.
I read this one before but was confused because I am so terrible with photoshop. I don't know how to make a background transparent but will figure it out because I may even like this way better if it works as well. (you could save different shapes for different situations. VS10 users couldn't do this though due to no moving paths.
Thanks for the info.
- Ron P.
- Advisor
- Posts: 12002
- Joined: Tue May 10, 2005 12:45 am
- System_Drive: C
- 32bit or 64bit: 64 Bit
- motherboard: Hewlett-Packard 2AF3 1.0
- processor: 3.40 gigahertz Intel Core i7-4770
- ram: 16GB
- Video Card: NVIDIA GeForce GTX 645
- sound_card: NVIDIA High Definition Audio
- Hard_Drive_Capacity: 4TB
- Monitor/Display Make & Model: 1-HP 27" IPS, 1-Sanyo 21" TV/Monitor
- Corel programs: VS5,8.9,10-X5,PSP9-X8,CDGS-9,X4,Painter
- Location: Kansas, USA
Ebone,
Your correct about VS10. However for grins I did try it in VS10+. It works, as long as there is limited movement, and you are limited to a rectangle. I even tried using a vignette filter to give it an elliptical shape and soften it, but then I lost the transparency with the rectangle used for the blurring..
To create a transparent image, which is a 32 bit image, create your ellipse, using an outline shape, no fill. Then save it to a format that supports transparency, such as PNG, GIF, TGA. The alpha channel should be recognized in MSP or VS.
Your correct about VS10. However for grins I did try it in VS10+. It works, as long as there is limited movement, and you are limited to a rectangle. I even tried using a vignette filter to give it an elliptical shape and soften it, but then I lost the transparency with the rectangle used for the blurring..
To create a transparent image, which is a 32 bit image, create your ellipse, using an outline shape, no fill. Then save it to a format that supports transparency, such as PNG, GIF, TGA. The alpha channel should be recognized in MSP or VS.
Ron Petersen, Web Board Administrator
-
sjj1805
- Posts: 14383
- Joined: Wed Jan 26, 2005 7:20 am
- System_Drive: C
- 32bit or 64bit: 32 Bit
- motherboard: Equium P200-178
- processor: Intel Pentium Dual-Core Processor T2080
- ram: 2 GB
- Video Card: Intel 945 Express
- sound_card: Intel GMA 950
- Hard_Drive_Capacity: 1160 GB
- Location: Birmingham UK
Here is an often overlooked alternative method built into VideoStudio:
MediaStudio/VideoStudio: Use A Video Sequence File
MediaStudio/VideoStudio: Use A Video Sequence File
alright guys, I'm lost and about to throw my cpu out the window.
I've tried following devil and vidoman's advice and can't get either to work. As simple as everyone makes it sound, I must be missing something very easy.
First of all it took me about 2 hours to get a transparent background to be transparent in MSP. (using photoshop to create). It seems that only png works for some reason. On all the others, the supposedly transparent area shows up as solid white.
If I make say a green ellipse and float the ellipse over someone's face. Add mosaic filter and then key out the green, then the filter is also keyed out. I even tried it with the inverse. I made everything outside of the inverse green and same result.
What am I missing here?!?!?
I've tried following devil and vidoman's advice and can't get either to work. As simple as everyone makes it sound, I must be missing something very easy.
First of all it took me about 2 hours to get a transparent background to be transparent in MSP. (using photoshop to create). It seems that only png works for some reason. On all the others, the supposedly transparent area shows up as solid white.
If I make say a green ellipse and float the ellipse over someone's face. Add mosaic filter and then key out the green, then the filter is also keyed out. I even tried it with the inverse. I made everything outside of the inverse green and same result.
What am I missing here?!?!?
Using this method, it doesn't matter if the ellipse shape has a black rectangle around it, so just use MS Paint to make a BMP file of a white circle/ellipse on a black background. You can also adjust the ellipse using your moving path, so don't worry too much about getting the shape right.
HTH
- 1. Put your face video on V1 and your ellipse image on V2 and float the ellipse over the face using a moving path.
2. When you're happy, disable V1 (click the "eye" shape on V1 to "close" it).
3. Render the result to an AVI - you'll have a white ellipse floating about over a black background.
4. Enable V1 and copy it to V3. Disable V2. Disable A3 if necessary.
5. In V3's overlay options, specify "gray key" using video matte, and point it at the AVI you just created. Select "invert overlay area" and (if you want) increase "soft edge" to the first notch.
6. In V3's video filters, add Gaussian Blur.
HTH
Just as a side note, all the above has a name, it is called "match moving".
In a nutshell, however you accomplish the effect, you have to match a bit mapped graphic over your target subject.
I found the easiest to be creating a bitmap file, i.e., a white circle the approximate size with a black background. (actually background really wont matter, but, in a way it does.) (bitmap could be .bmp, .jpg, your choice..)
I usually make the image file 3 times wider and 3 times taller than my video settings. This allows for path movements that may take it close to offscreen.
Using this image file as a video matte, I overlay it to the subject, then frame by frame, 2d basic path over the subject.
Once I am happy with the path's, add what ever filter, even a big yellow smiley face.
I said background won't matter, but, black easier when using it as a matte, but, really, depends on your source colors.
In a nutshell, however you accomplish the effect, you have to match a bit mapped graphic over your target subject.
I found the easiest to be creating a bitmap file, i.e., a white circle the approximate size with a black background. (actually background really wont matter, but, in a way it does.) (bitmap could be .bmp, .jpg, your choice..)
I usually make the image file 3 times wider and 3 times taller than my video settings. This allows for path movements that may take it close to offscreen.
Using this image file as a video matte, I overlay it to the subject, then frame by frame, 2d basic path over the subject.
Once I am happy with the path's, add what ever filter, even a big yellow smiley face.
I said background won't matter, but, black easier when using it as a matte, but, really, depends on your source colors.
It's not really anything to do with matchmoving! It's more a case of single-feature tracking.
Matchmoving is the process of matching the camera's movement to the scenery. The whole point is that you can composite 3D items or scenery onto the video realistically. It's a complicated mathematical process in seven dimensions (X-, Y-, Z-translation, roll, pitch, yaw, focal length) which usually needs specific software such as Icarus, boujou, PF Hoe, SynthEyes etc. to achieve well.
Single feature tracking is picking something from the video (such as a face) and following it around in two dimensions to blank it out. To do matchmoving well, it needs to be invisible to the viewer. The sort of 2D tracking the OP wants is (by definition) highly visible - it's obliterating a face by overtly artificial means.
Matchmoving is the process of matching the camera's movement to the scenery. The whole point is that you can composite 3D items or scenery onto the video realistically. It's a complicated mathematical process in seven dimensions (X-, Y-, Z-translation, roll, pitch, yaw, focal length) which usually needs specific software such as Icarus, boujou, PF Hoe, SynthEyes etc. to achieve well.
Single feature tracking is picking something from the video (such as a face) and following it around in two dimensions to blank it out. To do matchmoving well, it needs to be invisible to the viewer. The sort of 2D tracking the OP wants is (by definition) highly visible - it's obliterating a face by overtly artificial means.
Thank you so much for being so explanatory Gorf. I got it now.
Apparently, my stuff wouldn't work because I didn't know I had to render before "applying" it to the vid. When you're not a noob, you take for granted the steps that the noobophiles (like my self) do not know.
I did it exactly your way Gorf, and it worked perfectly. I did find that my preference was to do everything you said but use the green ellipse/transparent background image I had made. Instead of gray key, use chromakey... The only difference/advantage is that when you're working on the moving path, it's easier to use this one because there's no background that gets in your way of sight. It makes it a bit easier if there's lots of movement or you're trying to apply your overlay very precisely.
So if any other fellow noobs are reading this and are confused, I'll sum up my process.
*If you're using VideoStudio, go with the original process I explained above or the star filter. In VS, you're basically limited to these 2 due to no moving paths.
*If you're using MSP, then go with Gorf's method (and/or the other variations posted above.) They're much better.
What I did, step by step(mostly quoting Gorf's instructions):
1) Using photoshop/other image editor create a green ellipse on transparent background.
2) Put your face video on V1 and ellipse image on V2 and float the ellipse using 2d moving path.
3)"Close" V1 by clicking the "eye".
4) Render to AVI.
5) Enable V1 and copy to V3. Disable V2.
6) Add greenscreen overlay to V3. In overlay options, specify "bluescreen" using "video matte" and point to the AVI you created. Select "invert overlay area" and increase "soft edge" if desired.
7) Apply whatever filter you choose to V3(Gassian Blur, Mosaic,etc.)
Done.
BTW, I accidentally "discovered" a cool effect when I used this method and forgot to "invert overlay area." Everything was blurred (set to about 10 on gaussian blur) EXCEPT the face. It looked really cool and almost made the face stand out in 3D.
Thanks everyone for helping me with this.
Apparently, my stuff wouldn't work because I didn't know I had to render before "applying" it to the vid. When you're not a noob, you take for granted the steps that the noobophiles (like my self) do not know.
I did it exactly your way Gorf, and it worked perfectly. I did find that my preference was to do everything you said but use the green ellipse/transparent background image I had made. Instead of gray key, use chromakey... The only difference/advantage is that when you're working on the moving path, it's easier to use this one because there's no background that gets in your way of sight. It makes it a bit easier if there's lots of movement or you're trying to apply your overlay very precisely.
So if any other fellow noobs are reading this and are confused, I'll sum up my process.
*If you're using VideoStudio, go with the original process I explained above or the star filter. In VS, you're basically limited to these 2 due to no moving paths.
*If you're using MSP, then go with Gorf's method (and/or the other variations posted above.) They're much better.
What I did, step by step(mostly quoting Gorf's instructions):
1) Using photoshop/other image editor create a green ellipse on transparent background.
2) Put your face video on V1 and ellipse image on V2 and float the ellipse using 2d moving path.
3)"Close" V1 by clicking the "eye".
4) Render to AVI.
5) Enable V1 and copy to V3. Disable V2.
6) Add greenscreen overlay to V3. In overlay options, specify "bluescreen" using "video matte" and point to the AVI you created. Select "invert overlay area" and increase "soft edge" if desired.
7) Apply whatever filter you choose to V3(Gassian Blur, Mosaic,etc.)
Done.
BTW, I accidentally "discovered" a cool effect when I used this method and forgot to "invert overlay area." Everything was blurred (set to about 10 on gaussian blur) EXCEPT the face. It looked really cool and almost made the face stand out in 3D.
Thanks everyone for helping me with this.
As you've worked out for yourself: There isn't a single way of achieving the effect you want. This is so true for virtually every effect a user may be aiming for. With that in mind, my method (or yours) is no better (or worse) than any of the other solutions, above. It's just different.ebone wrote:<snip>*If you're using MSP, then go with Gorf's method (and/or the other variations posted above.) They're much better.
And here's where I make my confession. I never, ever remember whether I need "invert overlay area". There are some things you need to learn, and some you don't. This is one of the things you don't. It's what the preview window is for. You know it will work one way or the other - so you try one way and if that's not it, it must be the other.ebone wrote:BTW, I accidentally "discovered" a cool effect when I used this method and forgot to "invert overlay area." Everything was blurred (set to about 10 on gaussian blur) EXCEPT the face. It looked really cool and almost made the face stand out in 3D.
I know this thread is supposed to be applied to VS and MSP but one way not mentioned is using MSP's CG Infinity module. This way has no key frame limit and one can resize and rotate the object over their background with the use of keyframing. Simply, set the duration, frame rate and frame size to match your video. Load your desired file as background. Example: Original.avi. Use the shape tool and select shape. One can create a custom shape using the adjust tool to add nodes. Example: Create a circle into a binocular shape. Enable Object Style Panel and under General, select 0 for Line Width, 0 for Uniform Transparency. Under Color tab for Fill Color use eyedropper to select a single color. Example: Yellow. Line Color: None. Under Shadow tab select No shadow. Enable Time Control Panel and you will see the timeline that will allow you to set keys. Select Moving Path Tool to set start and end coordinates. Frame 1 set your object where you want it to start. Go to the end frame and drag E to where you want your object to end. One should have a straight line from S to E. If one needs to add precise keys for control of their object over the video, one needs to use the Key Frame Controller in the Time Control Panel and add keys. Then by selecting the Adjust Moving Path button, one can drag their object anywhere along the path for assigned positioning to set key frames. If one needs to alter the path, one can select the "+" to Add Node to Moving Path to change their path shape. Example: Z-shaped path. When satisfied with outcome of the object tracking over desired areas of the video, Save As .ucg project file.
Switch to MSP Editor module. V1 track will be the original video file. For V2 track, copy and paste original video and add your desired filter/s. Example: Mosaic. Now select Overlay Options. For Type select Alpha Channel and Mask select Video Matte. Browse to your saved .ucg file and apply. That's it. A perfect tracking matte that required no additional renders, color keying, etc... Save for final output.
Hope this helps and has shed some light on one of MSP's less talked about modules.
Switch to MSP Editor module. V1 track will be the original video file. For V2 track, copy and paste original video and add your desired filter/s. Example: Mosaic. Now select Overlay Options. For Type select Alpha Channel and Mask select Video Matte. Browse to your saved .ucg file and apply. That's it. A perfect tracking matte that required no additional renders, color keying, etc... Save for final output.
Hope this helps and has shed some light on one of MSP's less talked about modules.
Last edited by rwernyei on Wed Mar 21, 2007 5:14 am, edited 1 time in total.
To clear up any possible confusion CGI is not a part of MSP8 but was for MSP7 and earlier. If you don't have an earlier version, you can buy it for a moderate sum as a separate item, along with VideoPaint. I forget offhand the name of the add-on, but you can easily find it on the Ulead website.
[b][i][color=red]Devil[/color][/i][/b]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
