General Discussion

[DarkTl]

Found another renpy game in development, rogue-like evolution. It's more like a VN, but it uses a similar interaction menu, which means I can borrow some ideas and lines.

I found "cured of greed" version, but in fact the game hosts at patreon, and the developer makes 1000$ per month just for giving access to his work in progress. Unbelievable 

[Xela]

Oki, I've threw in some female clients for testing, you can increase the percentage if you need more. I think for beta I'll write a system that advertises based on target audience, so it should be possible to run a brothel mostly for female clients (if I decide to take that direction).

I've to go with this for BE GFX:

1) We will still have to calc pauses but Only for start of animation.
2) Each sequence will be moved it's own func/method and hiding or stuff between those animations will be put on new CPU threads.
3) At the end of show_gfx method we'll set all sprites to correct positions.

There are two issues that I can foresee but it is still less than with any other design. Also those issues are not guaranteed to cause problems and there are ways of solving them even if they do (although with clumsy code). There are many other ways but this seems the best so far, normally this would not even be a concern because in most BEs of similar complexity it is impossible to skip GFX, but we can prolly afford to allow that

[DarkTl]

Doesn't renpy use GPU for animations?

[DarkTl]

As for groups...
We could or could not have a certain number of people in the group action. So in some cases we know for sure how many people do it, and we even know their genders. And in some cases we can write anything, and it won't be contrary to the picture.
So we kinda generate customers groups depending on tags, not the other way around. Unlike all other acts.

[Xela]

Doesn't renpy use GPU for animations?

Yeah, but the problem is control, not animations themselves. Because we need a lot of stuff happening at very specific times, calculating the right pause before every next animation is really, really tedious (and getting close to impossible the more stuff we add to attacks + I wanna have even moar in the future ).

So since just pausing is getting too complicated, we need to manage the graphics show/hide stuff better, options as I see them:

1) Sim'Py.
- This can take over gameflow and manage every little thing separately at a cost of not being able to skip animations without too much extra coding. There are no limits here beside this one.

2) Timers.
- We add normal Ren'Py timers for delayed/conditioned execution of GFX events. We still have to use a lot of pauses because timers not actually stop game progression (by default, they need to be expanded for that with a while loop (clumsy code again)).

3) New CPU threads.
- CPU Threads will allow us to pause the flow of any particular animation sequence without interfering with the main thread. This means less pauses and somewhat better control.

4) So what Jake did, have each action request pause and have a general manager but his design seemed really linear unless I misunderstood something. In any case, this is something I'd prolly like to avoid.

===
I am going with a combination of 2 and 3 for now. First tests seem to run ok, I got a really weird Ren'Py error once but it never came back. I'll know if it's a final solution once I manage to find time to generalize GFX over all classes and see how new system manages the more complicated once.

As for groups...
We could or could not have a certain number of people in the group action. So in some cases we know for sure how many people do it, and we even know their genders. And in some cases we can write anything, and it won't be contrary to the picture.
So we kinda generate customers groups depending on tags, not the other way around. Unlike all other acts.

I have no idea how to handle that without "cheating" atm. Maybe it'll get better after I add "negotiation" with customers before Job is executed.

[DarkTl]

I have no idea how to handle that without "cheating" atm. Maybe it'll get better after I add "negotiation" with customers before Job is executed.

Why? Atm we check what kind of suitable tags combinations character has before picking one of them randomly and proceed with text and skills manipulations.

So for groups we check what kind of suitable groups pictures have all available characters in the current building before attaching customers, and act based on results of that check. Might be slow, but on the other hand groups events are supposed to be rare, well paid and exhausting.

[Xela]

Why? Atm we check what kind of suitable tags combinations character has before picking one of them randomly and proceed with text and skills manipulations.

We need to check before we call the main body of the job, client should come in with expectations and demands which can or cannot be realized by the business/workers.

So for groups we check what kind of suitable groups pictures have all available characters in the current building before attaching customers, and act based on results of that check. Might be slow, but on the other hand groups events are supposed to be rare, well paid and exhausting.

You see... that is f*cked up because we cannot have content of packs guiding logic progression in a Sim game, it the most ridiculous, hard to balance, sh!tty idea ever and will prolly be hard to compensate. For any act, we'll have to revert to default pics if the "main" act is missing. In Jobs them selves we can settle specifics but general acts have to be ran as long as there are clients/workers and business generally permits it.

I am very much against basing a Sim game on unpredictable factors like images/tags that we have 0 control of. Our usual approach of running the code trying to find the best pic and reverting to some default if none is found is much better.

[DarkTl]

Speaking of reverting to default pics, the game doesn't do it by itself. Recently I tried to show not nude rest picture, and one of small random packs didn't have one. So it showed me a huge "no image found" picture instead, which is really not an option as long as we have at least one profile picture.

I'll try to find some kind of default replacers for all actions that won't show character directly, thus will be suitable every time.
But at very least a random profile picture is better than that huge black message in any case I can think of.

I am very much against basing a Sim game on unpredictable factors like images/tags that we have 0 control of.

Another solution is to add groups as a special type (maybe even gender) of customers, that are rare and have much more money than single customers. But don't have a specific number of members when we don't want it.

[Xela]

No idea, there are prolly some special rules in play there, you should have gotten the profile pic...

Edit: Logical targeting seems to be messed up prolly because I moved the calc methods, need to check that sh!t...

[DarkTl]

I happened to notice that if min stat becomes > max stat, the stat becomes 0/0. It's not an issue, but I wonder if it's intentional.

[Xela]

I happened to notice that if min stat becomes > max stat, the stat becomes 0/0. It's not an issue, but I wonder if it's intentional.

Yes, that was intended (not exactly like that but still).

[DarkTl]

Possibly decent ideas from that renpy game above:

- Character can be submissive or self-confident. Submissive ones need less disposition to do stuff, but usually don't act on their own. Self-confident ones have higher checks, but can propose MC to do something (like another pose) instead of simply follow his orders.
This system uses two stats for some reason, in our case it's just character stat.

- Girls are limited only by vitality, but MC has a limited amount of libido tied to, well, capacity. No liquid = no cumming, kinda makes sense.
Maximum can be upgraded there via traits, value restores over time.

- It has drugs system. It uses two stats, one shows how quickly increases the need for fix over time, another one shows the current need for fix. The more the need, the more character is willing to do to reduce it.

[Xela]

Possibly decent ideas from that renpy game above:

- Character can be submissive or self-confident. Submissive ones need less disposition to do stuff, but usually don't act on their own. Self-confident ones have higher checks, but can propose MC to do something (like another pose) instead of simply follow his orders.
This system uses two stats for some reason, in our case it's just character stat.

Can you add a link? I couldn't find it by name.

We have traits for this, paired with character stat (through automatic normalization system), it will do, possibly better.

- Girls are limited only by vitality, but MC has a limited amount of libido tied to, well, capacity. No liquid = no cumming, kinda makes sense.
Maximum can be upgraded there via traits, value restores over time.

It's a fairly simple feature we can add at any time.

- It has drugs system. It uses two stats, one shows how quickly increases the need for fix over time, another one shows the current need for fix. The more the need, the more character is willing to do to reduce it.

We talked about something like that but it's not that much of a priority either.

[DarkTl]

Can you add a link? I couldn't find it by name.

Do you have an account at pornolab? If not, time to make it 
The game doesn't have much except good ideas and original art/lines though. I seriously can't believe that someone can achieve so little for two years of development and get away with it while getting 1000$ per month 

We talked about something like that but it's not that much of a priority either.

We never discusses possible concepts though. The idea is more interesting than in wm for instance.

[Xela]

...

Primitive gameplay + consistent art. That always seems to work well.

We never discusses possible concepts though. The idea is more interesting than in wm for instance.

We have and I thought that through a lot myself but it's not a priority, we don't even discuss jobs much which are far more important