Yes, the Yellow Tang is in the tank. Now the Wimplefish, Achilles Tang, and Powder Blue Tang are, too. But last night I noticed a small problem. The image quality of the fish just didn't look quite as good as they did in the old program, so I did a couple of screen-captures and compared the two. It turns out that the fish are about 10% smaller in the new version. This is because the camera is slightly farther away. I use 128-pixel-wide textures for most of the fish because that's about how big they end up on the screen, but the Yellow Tang is only about 100 pixels wide now. Without going into the details of MIPmapping, let's just say that some of the detail in the texture never makes it onto the screen, resulting in a slightly ragged image.

So, I have to make a decision from among these possibilites:

1. Use the current system, and just live with the lower image quality for now. All the fish models will be replaced soon with a whole new system, but I promised Encore I'd get a Beta out soon, and coming up with a whole new fish-object system was not on the schedule yet.

2. Enlarge the mesh for each fish as it's loaded. I could write a routine to scale the vertices without adding too much to the loading times, but behavior and collision avoidance might be adversely affected.

3. Go back to Lightwave and recreate each fish, but larger. Very time consuming, and I no longer have some of the programs I used to convert the files from Lightwave to Direct-X while retaining the individual parts of the mesh.

4. Change the camera position so that the fish appear to be the same size that they used to be. The current position is the result of many days of trial-and-error to get the optimum image quality of the background objects without any "behind the scenes" elements showing as the camera pans back and forth.

I'll do some quick experiments regarding #4, but if those fail, I'm leaning toward #1.

I might lean towards #1 to start with. Assuming that you do not have every single object already placed in the tank as of yet, if you change the camera then will you not have to go back later and redo slightly any objects placed in the tank from the camera movement onwards?

If thats correct, I would leave the camera where it is, and save the "later tank object rework time" for the fish which you already planned to redo anyhow.

Same here! .... If, "All the fish models will be replaced soon with a whole new system", then it sounds like you would be making additional unnecessary work for yourself, if you adopt one of the other options. ... And compromising collision avoidance, or optimum image quality of the background objects, doesn't sound like something J Sachs would do! ;)

My quick experiments with the camera position didn't work. Gaps and seams started showing up in the tank objects and while the fish were the right size, their relative scale compared with other tank objects was still too small. So, I guess I'll go with Option 1.

Funny, I came here to ask that very same question (about the Yellow Tang, that is). :)

#1 sounds like the proper way to go - heck it's supposed to be a *beta*, right? (but #3 sounded very Jim... ;) )

Jim,
There is another option, change the zoom of the camera so the fish will be to the size you want them. Since the direction of the view of the camera did not change, it should still hide the other background as expected.
The weird part of this mode is the feeling of being closer than you really are to the fish.

I tried that as part of the set of experiments I did earlier today. It helped in some ways, but the relative size of the fish is still a major problem. The background objects, especially things like clams and barnacles are just too big compared to the fish.

I decided to see what it would actually take to really do the fish right at this point instead of later. I'll be spending the evening re-creating one of the fish from scratch as a single object, recording all the important vertices (eyes, fin tips, etc.), and saving it as a DirectX file. Maybe I've been dreading this so long that my mental image of the task is blown all out of proportion.

Your idea of scaling the fish once loaded should also work. You just have to make sure you scale your collision boxes or sphere the same amount.
Basically need to apply the same transformations to the fish and the collision models that you decide to use.

Collision boxes or spheres - wouldn't those be nice? I hope to include those into 3.0, but for now each fish just has a central point which tries to sense the central points of other fish and keep them an appropriate distance apart.

That is pretty much a sphere with an expanding radius. :)

Even if this takes a lot of time, maybe this is the right way to go, instead of spending time on adding the old outdated ones.
Please let us know how this turns out! Hopefully there can be a beta once a few fish are re-created. Will be so cool to have a new version, and then regular updates as soon as a new fish is created.

I just finished the basic model for a prototype fish in Lightwave. It was just as hard as I thought it would be, trying to anticipate everything the critter would need to do. All the parts which used to be separate objects (eyes, side fins, low fins) are now part of one single mesh. In addition, the mouth will be able to move, and the gills work. But I've got to send in an update tomorrow, and have nothing to show, so I'll have to put in all the old fish for now. If I pull an all-nighter I should just be able to make it.

Poor Jim. Don't overdo!

What's going on with Encore? Isn't Prolific your publisher?

Jim's working against deadlines?! Wow!

Why are you up Minn? It must be about three AM there, I just got up about an hour and a half ago. I wanted to write on MC a bit.