Jump to content

How to make movies out of reference photos.


Recommended Posts

This is a neat little trick I discovered today. If you're like me, and you're using stock photos to make your tulpa, you can turn those stock photos into a video of your tulpa actually turning his or her head and changing facial expressions. You'll need maybe fifty or more photos. I'm working with over 200.


I use Linux as my operating system, and I'm sad to say that this doesn't work on it. It will work on Windows XP and later, and I'm pretty sure it works on Mac too. (Well, you can get it working in Linux with Wine, but it may be more work than it's worth if your Wine isn't configured.)


What you need to do is download Picasa if you don't already have it, or update Picasa if you have an old version -- you need 3.8 or later.


Import your tulpa's pictures (and only your tulpa's pictures, if you can manage it -- if not, it'll take a little longer, but no big deal). Now tag all the pictures of your tulpa with his or her name. Picasa will probably leave a few pictures out because it doesn't recognize the face in them. That's ok, just forget about those pictures for now.


Picasa will make a category in the library called "People" and under that will be your tulpa's name. Click on the name, and select all the photos. You need to be in the People ... Tulpa category -- any other category won't work. The feature is practically hidden. If you're in the right place, a new button will appear in the center pane above the pictures. It'll appear next to the "create movie presentation" button, and it looks just like it, except it also has the "people" icon overlaid on it. Click that button.


Now Picasa takes you to the Movie Maker tab. The Options will be shown to the left. Select "Best Transitions", drag the top of the two sliders all the way to the right, uncheck "Remove Low Resolution Faces" and then click "Apply". With my collection, this convinces Picasa to use 187 photos instead of four. It's still not using all 229 photos, but that's ok -- 187 is plenty.


Click on the "Movie" tab up above the "Options" pane. Change "Transition Style" to "Dissolve". I like 1 or 2 second Slide Duration, Overlap 100%, and 320x240 in Dimensions, so try that out and click the green play button at the bottom.


Picasa will play a movie for you now. This is Picasa's idea of what looks best with the photos you selected. It won't be perfect, but you'll probably get some beautiful transitions where for five or six photos, it looks like a seamless movie of your tulpa moving. Cut out the stuff that doesn't transition too well, and tweak the rest.


Sometimes Picasa makes things look a little ugly because it's trying to use every single photo exactly once. You can make things look a lot better by not using some photos, and by using other photos several times. Picasa allows you to drag photos left or right on the timeline, and delete them, but you can be sneaky by copying photos and importing the copies, and by deselecting the photos that don't work well before you hit the face movie button.


I like to remove the title slide, and make the last photo a copy of the first one, so the movie loops seamlessly.


Play around with it. Tweak the movie options, make multiple movies, and string them together with Windows Movie Maker, or whatever you Mac people use. Picasa does not make this an easy tool to use (no undo, no copy & paste, no import), but you can get some really beautiful results if you take the time.


Sorry, no example yet. I'm still working on mine, so I was going to link a video from youtube, but all the ones on youtube are of kids growing up, except for the one of that one guy who stood there for like an hour with exactly the same facial expression while people took pictures with him. Anyway, while you can see face movies on youtube, none of them show what I'm doing with it.


I'll upload one soon if nobody beats me to it.

Link to comment
Share on other sites

  • 2 weeks later...

I know, it took me forever to get this example up. I'm still not completely happy with it, but here's a brief segment. I wanted to highlight the wink, but I think the mouth animation came out looking better.


Ideally, I'll do a little photo editing before the final video, so I can get all the images to look roughly the same quality, not have makeup and distracting backgrounds fading in and out of existence. But at this point, I'm just figuring out what pictures work well together. And this project is doing wonders for my visualization.

Link to comment
Share on other sites

I start with Shutterstock.com, because when you're looking at a picture, it shows you others of the same model. If you want even more than Shutterstock has, often you can find a few more by googling the photographer's user name -- sometimes they like to post different photos of the same model on different websites.

Link to comment
Share on other sites

Watermarks are hard. The easiest way to get rid of them is to buy the photos. I'm thinking I might celebrate my tulpa's first birthday in six months by getting a month Shutterstock subscription. But that's a lot of money, and if you're not sure she's going to stay in that form, or you don't have a credit card, or you're broke, or whatever, I can understand not wanting to pay it. So here are some free methods.


There are several guides on Google for doing this with various programs, but many of them are hard or don't work very well.


You can do a reverse image search, where google will look for images that look similar to the one you're searching for, and it might find the image on somebody's site without the watermark -- this is probably the best option, but it requires that somebody uses the image where google can see it.


When you can't find the photo without the watermark, often you can find the photo with the watermark in different places. You can use a simple image editor like MS Paint to copy the parts of one photo without the watermark over another photo with the watermark. That can help get rid of a lot of it, but when the watermarks overlap, there's nothing you can do about it.


Personally, I usually just ignore the watermark. In my example video, all eight of those pictures still have the watermark -- it's just that the watermark never covered the face, so it got cropped out of the video. (You can see a little bit of the diagonal line in the 7th picture though.)

Link to comment
Share on other sites

  • 2 weeks later...

One more tip that's worked really well for me: sort all your images by what direction the model is facing. If they're facing left, put that in a folder marked left, if they're facing right, put it in a folder marked right. If they're facing forward, try really hard to see if it's slightly one way or the other. I think the easiest way to tell is by ears -- if you can see one ear and not the other, or you think you could see one ear if there weren't hair or a hat or whatever in the way, then your model is facing away from that ear.


Once all your photos have been sorted like this, take all the photos facing one direction, and flip them so they're facing the other way. In linux, you can do this from the command line with ImageMagick. I found a windows program that does it too, but it came packaged with adware. And if you really need to, you can do it by hand with the windows image viewer or Paint -- it just takes a lot longer.


When all the pictures are facing forward or to one side, the video comes out a whole lot better, and transitions are much smoother. It's really worth the effort.


As for watermarks, I just found a program that helps: Image Stacker (Windows only). It combines similar images. So if you have two versions of an image with different watermarks, it can combine them into one image with a watermark only where the watermarks overlap. It's very particular though -- if you have 16 bit images or higher, it apparently doesn't work. It also won't accept if the images are different sizes. And if the images are too big, you need to register the program before it will work with them. And it only tells you all these things in a command prompt window that flashes past in an instant, so it's very hard to tell what's wrong.


Shutterstock adds 20 (I think) pixels to the bottom of an image, which need to be removed in MS Paint or something before Image Stacker will combine it with an image from Dreamstime or whatever. And if you find images of different sizes with reverse image search, you'll have to stretch/skew the smaller ones in MS Paint until they're exactly the same size as the larger ones.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Create New...