Messages

Hey, I tried this awhile back. Here are some tips for you, accompanied by some WiP images that I never really finished.

I started this walk cycle draft using a body I'd made in Poser, and picked out I think 12 different keyframes on which to base my animation. I copied + pasted into Illustrator, and started by drawing each part individually over the Poser model. I used the pen tool and the mouse, manually setting my curves. I over-drew the joints, so that there was overlap between parts. I drew the head and hat really large and rough with a tablet in Photoshop, and Live Traced them in Illustrator, because this originally just started as an attempt to do colored linework:

Were I to do this again (there will be a lot of "Were I to do this again" in this post), I'd either do it all in Illustrator, or leave my vectors headless before moving them over to Photoshop and I'd draw the heads there. At this point, I had all the pieces you need to make a police detective:



I was working through this really quickly, so there's some weird stuff, like the front ball joint of the shoe that has a weird trailing point. I copied the hands over from another failed walk cycle, so they don't quite fit with the rest of the body.

I kept each of my walk cycle guides on a separate layer. I did a simple print screen, and used the top of the GUI window to make sure they were aligned. I'd show an image for this, but I deleted them all from the .ai file, apparently. I started by placing all of the parts on the model and arranging them for the first frame. Then, I'd select all, and ctrl+C to copy the parts, hide and lock that layer, then move up to the next one and ctrl+F (to paste in front, in the same place). This meant I wasn't starting from scratch on every frame, but was able to move the parts just slightly from one frame to the next.

I pretty much just selected each part, rotated, moved and scaled as I needed to. Were I to do this again, I would use the following method:

1. Establish a circleof a few pixels around which each part should rotate. For example, in making the arm, put a circle in the center of the shoulder socket, one at the elbow, one at the wrist.

2. Use alt-click with the rotation tool to select these circles (one at a time, see below), which will define the point around which the object is rotated.

2. Start by rotating the whole arm, then the forearm, then the hand. Work from the joint that moves the most amount of mass to the one that moves the smallest.

...This would give more consistency than what I did, which was to just kind of eyeball it.

I did this 12 times, until I had a full walkcycle. Some parts needed to be changed a little, especially around the knees. Were I to do this again, I might make a separate, more consistent knee and elbow piece to cover that joint. There are, on this version, no elbows of which to speak. I left all of the parts unattached, just kind of hanging out there, looking like this:



To move everything over to Photoshop, I drew a printer's mark bullseye on a separate layer. I then went in to each individual layer and combined the parts that needed to be combined -- the upper arm and forearm, the thigh and calf, the two parts of the shoe -- using the Pathfinder unite function. Generally, this worked well, and kept the same color + stroke consistent. Sometimes I'd adjust a few things afterward, but not too often.

Then, I'd copy each frame over to Photoshop, resize a little, and line up the bullseyes, again keeping each frame in a separate layer. I tried to start with my widest frame, where the feet were furthest apart, but I think I still wound up using the expand canvas function to make sure everything fit. With all layers visible, it looks like this:



Animated, it looked something like this:



This walk, because I'd made some "toughness" adjustments in Poser, was a bit overblown for me. But it was a cinch to go back and make the motions more subtle for the next version, though I didn't unite my segments here (because I was still just experimenting [and ignore that weird half-step -- obviously a mistake that I should've corrected!]):



It took me, from start to finish, Poser, drawing the head and all, about two hours to do the first walk, and another 20 minutes to adjust for the second walk. Honestly, making the animated gif took up most of that time.

The next day, I wondered how I would fare if I just kind of eyeballed the movements for an animation. This one is a rough of a quick draw and two gunshots fired at a speeding car. Imagine that he has two hands, that the hands are also animated, and that there's a gun. I have faith in your imagination! I was too lazy / still experimenting to draw them myself!



It turned out pretty well, I thought! Only 20 minutes to do this one.

The big con to this method is foreshortening. It works swell for east and west views, but you'd have to change the scale of the arms and legs for north to south, and your diagonals. Also, any motion that made the sprite turn from one view to another would be tricky to line up correctly. My attention span wandered before I got that far with it, however.

I hope this is somehow useful, even if you just learn from my mistakes. I expect others will chime in to say, "WHY IN THE WORLD WOULD YOU DO THAT THIS WAY WHEN [simple way] WORKS BETTER?" And the answer will be because I didn't know about that way, and thanks for telling me (and slasher) that it existed.

Once you know what to look for, it becomes blatantly apparent. Sad indeed.

I'm pretty sure I eventually, sadly, figured it out. But my first guess, based on Calin's clue, was someone who got massive press elsewhere that led one-time posters to their thread. I agree we shouldn't be speculating in public, but maybe it's best if we don't speculate in private either!

Look for game threads with unfamiliar posters who have posted almost soley on those threads.

Might be worth naming and shaming at this point so we don't start hunting witches.

If you take a closer look at the video you shared, it's kind of doing just that. Those ripples roll to and fro, but they have a separate horizontal movement.

I've never seen Victor Surge, who created the first Slender Man images on Something Awful, reference Chzo. The character he used in the first photo-manipulations was taken from the 1979 movie Phantasm and pretty much the rest of the Slender Man mythos was, as Barthes writes, "Drawn from the innumerable centres of culture." That is to say that it's likely that Chzo, Phantasm, and Slender Man are all developed from fears and images shared by more than just their creators.

Thank you all so much for contributing to this thread. If no one else learns anything from it, I have gained much to think about in my art process.

Andail and Ascovel -- Would either of you be willing to make a process post showing your work from Sketch-Up to in-game? I've tried working from a CAD program, Chief Architect, for my indoors scenes, but find the perspective and camera placement to not function in a usable way for me. Andail, I especially liked the backgrounds in the Samaritan's Paradox demo. They reminded me a bit of the old XIII game, in that they hit a balance between a 2D and 3D look, sort of cel-shaded but better done than most stuff like that I've seen.

Ilych - I've said elsewhere that your game is one of the most beautiful I've ever played. Any chance you have one of your background files in layers so we can see your process as well? And you say you prefer the two-point perspective -- do you have thoughts on either the scaling issues that come with that, or mine and Technocrat's similar experience in being drawn to two-point for outside scenes and one-point for indoors? (Also what is that futuristic neo-classical background for? Looks good!)

Ben and 90s (together, 394) - Would you suggest that an adherence to perspective creates a certain kind of expectation -- verisimilitude in BG design = reality across all aspects of gameplay -- whereas the perspectiveless (or loose perspective) background artist throws the player off-kilter from the start? (And maybe Andail -- how do you see this affecting the play of your game, which seems to take place in two worlds [the real world, and the world of the missing novel?]?)

Thanks all!

If you're planning on going commercial, you might want to adjust the face so it's a little less Bronson, lest you be beset by litigious descendants. I might make that cross white like the collar, and maybe the reflections on the gun barrel too, to make them pop from behind the blood.

While you're going full-kitsch, have you considered playing off of the cross image with either of the capital T's found in "THE PRIEST"?

I suggest we put all of this into two boxes, stack the boxes, and carry them to the room where Emma is.

I find 3D programs incredibly frustrating, just because I haven't sat down to properly learn them. I'm hoping that continuing to work with the image will make it look less like the 2D texture on a 3D object, and I need to be more considerate of which things don't exist on the flat surfaces I've drawn -- like that "OPEN" sign throws that whole plane off. I should draw it in separately, thinking about how it fits in that space.

If nothing else, this will help me plot where things go. The trouble I have with drawing in perspective isn't so much the getting the parallel lines to align, it's with correctly plotting the distances between objects -- especially those that are evenly spaced. This kind of stuff, even using the formulas seen here, I usually get wrong:



And I think that I can make it work -- after all, I'm always drawing in a 2D environment.

I've just walked around PISS for a bit, and none of the backgrounds seem especially jarring. Are you still kind of eyeballing the vanishing points, or just winging everything? You might just have a better knack for it than I do!

I've always stuck to having a vanishing point on a layer, lines towards the VP on another layer, and drawing on the third. However, to make a nice-looking (non-squashed) background, it's a good idea to keep the VP outside of the area that's being drawn, when doing two-points.

I always try to do this, but I always wind up screwing up and drawing a perspective line on the wrong layer at least once and have to ctrl-Z my way out of it. Putting the vanishing points, which is a word I somehow forgot to write last night in making the original post, outside of the actual image is why I make a super wide image to crop down later. I know you can set the ruler guides outside of the canvas in Photoshop, but I like being able to see the perspective line all the way through.

I seem to have unconsciously gotten into the pattern of using one-point when doing an indoors scene, two point outdoors, and three if looking upwards at something really tall!

I understand the three-point, as that's a perfect use for it, but I feel drawn to the same one-point for indoors, two-point for out that you use. Why do you think that is? Are there some inherent benefits to that breakdown?

Now we will probably only need a flathead wherever we go.

[Citizen Kane]

Hi guys,

There's a wonderful documentary about the director John Ford where Steven Spielberg recounts a tale of meeting the director when Spielberg was a young man. I'm taking this transcript from the Pop Matters review of the doc:

Spielberg says that he first met Ford when he was only about 15, aspiring to be make movies like those he admired by Ford. “So you wanna be a picture maker?” he remembers Ford saying (Ford in his office, dressed like he had just returned from a safari instead of lunch). “What do you know about art?” He sent the boy to a wall in his office where he had hung a series of Western landscape paintings. Asking young Spielberg to identify the location of the horizon line in a couple of them, Ford pronounced, “When you can decide that putting the horizon at the top of the frame or the bottom of the frame is better than putting it in the middle of the frame, you may, someday, make a good picture maker. Now get outta here.” Spielberg smiles.

Now I understand a bit about horizon lines and perspective in film (Citizen Kane is a master course in cinematography, especially any scene that's a conversation between Jed Leland and C.F. Kane in a mostly empty newspaper office), and in still drawings. But video games are a different animal, especially 2D video games like ours where a severe change in perspective can necessitate redrawing sprites from different views. I was working on a technique for making backgrounds tonight and it occured to me to ask, if you'll excuse the goofy wordplay, your perspectives on perspective.

For instance, here are some questions I've got:

• How do you all generally manage perspective when designing your backgrounds (and, accordingly, your sprites)?
• Do you prefer one-point or two-point perspective (or, and I will be impressed with you if you answer this, three-point), and what do you see as the benefits and drawbacks of each?
• Do certain perspectives work better for interiors and others for exteriors? For specific environments?
• Is it jarring at all to mix one- and two-point perspective in the same game, the same area, the same room?
• Should the point in a one-point perspective image always be in the center? And how do you handle one-point perspective in a scrolling background?
• How do you handle horizon lines? Do you like them high or low? And if you like 'em low, how do you avoid drawing ceilings?
• How far can you break perspective for effect (ala Day of the Tentacle)? How and when do you know to do this?
• What techniques do you use in which programs to set up your perspective lines and points?

I'm sure there are many others out there. So I thought I'd open a forum post and invite you all to respond. Feel free to share your own work, backgrounds from games you like, crude mock-ups, or whatever. Share best practices, raise objections, offer philosophical ideas on the nature of art, or whatever you feel like doing. The general topic is perspective. Consider it open.



I'll start by sharing a technique I was testing tonight. I'm dealing with a lot of urban spaces in a game I'm planning, and got tired of drawing windows, doorframes, sidewalks, etc., in perspective all the time. So I cranked up Illustrator, and made a flat version of a building (This is a draft, just enough done to get a nice structure to work with. Also, this is scaled down from the giant file I accidentally made):



I then made a Photoshop file, 640x480 at first, and colored the thing neon green. Then, I expanded the canvas to 300% with the live green section in the center. I used Photoshop's ruler guides to set up a two-point perspective, pretty much at random distances, and a horizon line just a little north of center. I imported the flat images in chunks (the main face of the building, the two faces of the part of the building that protrudes onto the sidewalk, and the bits that make up the canopy), and used the Free Transform / Distort function to adjust them along my perspective lines. I very, very quickly threw a few shadows here and there, and some light. The perspective's still not perfect because I was trying to hurry, and the shadows were a quick inquiry into whether I could paint-over the distorted flat images to give a sense of depth (I think I need to invest in a more extensive test). Then I drew a quick...background to the background.



There are some problems that I think would have been solved had I kept my images Smart Objects -- the problems with the strip pattern on the canopy, for instance. I think there's some promise in this technique -- I could draw the same building from different views fairly easily if the game called for it, I'm able to standardize things like windows (which I think I can individualize later by painting over) and bricks, and, interestingly for this project, which is set in the town in which I live, I could make my own 3D models of the buildings. There are other things I'll have to through and adjust -- the open sign looks really flat, for instance.



Thoughts on what I've done here, or on perspective in general? I look forward to hearing what you all have to say.

For the most part, I agree with #1 and #3, but I'd exclude Chris Cooper's appearance in The Muppets from this plea.

My flag boy says to your flag boy "I'm gonna set your flag on fire."

[embed=425,349]http://www.youtube.com/watch?v=5ov37eJcuZE[/embed]
[embed=425,349]http://www.youtube.com/watch?v=8K9PN4xJEN4[/embed]

Iko iko an day...

Oh! Well, that's foolish. I'll have to try another type system then. Thanks for the clarification, and sorry for the miscommunication on my end.

Oh, I didn't know that you could only import bitmaps for those. That might explain it. I was trying to import the outline and fill of Cassanet, which was surely TrueType, for the speech dialogue. The outline font seemed to be off to the right about 4 pixels or so.

There aren't any fonts that come with outlines.

Not totally true, but I've tried using font systems that come with both outlines and fill faces in AGS, and I've never managed to make it work / line up / appear satisfactorily.

Thanks for this.

No, we're talking about the person who started this thread.

Well, hell. I already planned mine according to the rules!