These are super cool and creative maps. For anyone reading this, if you haven't checked them out, you really should. It's a blast.

A bit of a cross-post, but I thought others might find this interesting. It's an AI tool that generates tileable material images. It can take in a reference image and/or prompt and has various configuration settings.

Use the keyword "seamless" to generate images that tile in all 4 directions.

"neon" and "cyberpunk" as qualifiers can net you some neat results as well.

I've noticed a lot of the output quality is, unsurprisingly, how well you can "zero in" which terms are well-represented inside the training set. You can feed the same image back into the tool multiple times to refine the results.

The best way I've found tends to follow the pattern:

1. Play around with terms to get something vaguely like what you want. Settle on a best candidate.
2. Pass the best candidate image back in to the tool and try more qualifiers to "hone in" on what you want.
3. Repeat until you are happy with the result.

This, like any tool, requires some learning through experience to use most effectively. But even so it's pretty neat for what it can do.

replicate.com/tommoore515/material_stable_diffusion

• Can't get 100% secrets (sector bleed-over, unfinished areas, etc.)
• Secret is empty because the items inside have specific skill flags set
• Inescapable damaging floor pits, unless it's used for a puzzle/restart mechanic or in a careful way.

Agree with @Gibbon on platforming sections as well.

Not treasure per se but I did find a real-life secret passage once.

I was in an old house and I swore I felt a draft behind a bookcase that was built into the wall. When I looked around the edges, I noticed that the frame didn't quite touch the wall. With some nudging and wiggling, the entire bookcase slid free!

Behind it was a large open space with musty wood beams and thick dust on the floor. No treasure, sadly. Just dust and spiders.

For me, it is a tie between a well-executed "finale boss" map (the DBP with custom bosses was a great example of this), or a story-concluding map with exposition.

"Filler maps" in general is an insulting term. Its origins are from the music industry where hit songs would need to be accompanied by additional tracks in order to fill out an entire album, so low-quality, unoriginal or uninspired works were quickly put together to "fill in" those gaps, hence they were referred to as "filler." I consider the term always negative given what it means.

Pacing for a full map set is important, of course: "breather maps" have an clear place. Long maps need to be balanced by short maps, heavy action by quieter areas, etc. Without this the player will become overwhelmed.

This is close to possible already. The latest nVidia voice cloning requires only about 30 minutes of targeted training. The output is still a bit stiff compared to natural voices or trained voice clones with a richer training set. But, the technology is only going to improve (and I suspect, rapidly).

SLADE, copy/paste, "add to TEXTUREx" using the right-click menu. Rename from there, usually I try to group things using some kind of simple system, using a leading number or prefix.

Sprites and flats obviously are much easier; just paste them in and rename.

DoomTools is probably a better way, but I haven't bothered to learn it.

For DeHackEd, I just do it in a text editor. With BEX format, I find things are plain-language enough to not require any special tools.

I did have one opportunity to play it, and Doom VFR as well.

It's nothing like what you expect, in ways both good and bad.

One giant misconception I found is the thinking that VR is just video games but full 3D. This is such a massive misunderstanding that I did not realize until I was playing.

In real life, you move around in a 3D space with your body, and your vision is a narrow circle in the middle of your field of view. In a typical computer game FPS, that view is filled with your monitor and you get a clear image of the whole space. You move around with tiny finger movements. Your aim point or crosshair is fixed in the middle of your view point, and that spot is truly fixed. It never changes or moves around.

In VR, your vision is tracking your head, but you can only really focus on the middle of your view, and only while it's relatively static. This is far more disorienting than you think.

Your body is rigidly stuck in the middle of a "playing area" but you can wave your hands around to aim at things. Your aim point is where your hand is oriented plus where your wrist is, and so on. You took for granted that aiming at things is a given and now it requires your whole body to orient toward something and get your hand in just the right spot, and pull the triggers without moving your hand, and so on.

Movement is absolutely a different thing in VR. Your movement is based on tiny thumb-sticks and trying to aim while moving is, while not impossible, orders of magnitude harder.

The big thing to understand is that a game that is "awesome" on a computer or console is not automatically "double awesome" in VR. They are different media entirely. It's a bit like how video games don't usually translate to movies and vice-versa. They are different things with different constraints and objectives.

It's still a cool experience. The visuals are incredible and I also got to experience some legacy Doom maps in VR.

It was best to enjoy such things with monsters turned off so you could simply enjoy the views.

Ooh, I like this question. Neat idea.

For me, COMPBLUE / COMPSPAN / COMPTILE. Probably some COMPUTE1/COMP2 sprinkled in too.

SLADE, copy/paste, "add to TEXTUREx" using the right-click menu. Rename from there, usually I try to group things using some kind of simple system, using a leading number or prefix.

Sprites and flats obviously are much easier; just paste them in and rename.

DoomTools is probably a better way, but I haven't bothered to learn it.

For DeHackEd, I just do it in a text editor. With BEX format, I find things are plain-language enough to not require any special tools.

I don't think so. I probably would have had to say 'hi' or something.

I did have one opportunity to play it, and Doom VFR as well.

It's nothing like what you expect, in ways both good and bad.

One giant misconception I found is the thinking that VR is just video games but full 3D. This is such a massive misunderstanding that I did not realize until I was playing.

In real life, you move around in a 3D space with your body, and your vision is a narrow circle in the middle of your field of view. In a typical computer game FPS, that view is filled with your monitor and you get a clear image of the whole space. You move around with tiny finger movements. Your aim point or crosshair is fixed in the middle of your view point, and that spot is truly fixed. It never changes or moves around.

In VR, your vision is tracking your head, but you can only really focus on the middle of your view, and only while it's relatively static. This is far more disorienting than you think.

Your body is rigidly stuck in the middle of a "playing area" but you can wave your hands around to aim at things. Your aim point is where your hand is oriented plus where your wrist is, and so on. You took for granted that aiming at things is a given and now it requires your whole body to orient toward something and get your hand in just the right spot, and pull the triggers without moving your hand, and so on.

Movement is absolutely a different thing in VR. Your movement is based on tiny thumb-sticks and trying to aim while moving is, while not impossible, orders of magnitude harder.

The big thing to understand is that a game that is "awesome" on a computer or console is not automatically "double awesome" in VR. They are different media entirely. It's a bit like how video games don't usually translate to movies and vice-versa. They are different things with different constraints and objectives.

It's still a cool experience. The visuals are incredible and I also got to experience some legacy Doom maps in VR.

It was best to enjoy such things with monsters turned off so you could simply enjoy the views.

This is close to possible already. The latest nVidia voice cloning requires only about 30 minutes of targeted training. The output is still a bit stiff compared to natural voices or trained voice clones with a richer training set. But, the technology is only going to improve (and I suspect, rapidly).

Baron - Reduce HP by ~20-25%, add splash damage to fireballs.
Cyberdemon - No change except might add a health indicator of some kind, or a palette swap at 50%.
Spider Mastermind - Reduce physical size by a lot (up to 50%), fix blockmap issues so it doesn't get stuck on literally everything, decrease pain chance, increase hitscan damage but limit to 3-shot burst, add flamethrower AOE attack if the player gets too close. Forces the player to keep distance but stay in cover. Needs the most overhaul.
Icon of Sin - Move to a phased encounter. Phase ends are transitioned by via special attack cube(s) that teleport all players to the next phase encounter. Phase 1: cube spawns are slower, limited to Episode 1 bestiary (minus Baron of Hell). Phase 2: Cube spawns now slightly faster and include some attacks, like an explosive cube, or a cube that slows player movement momentarily if not dodged. Phase 3: Normal dual-spawn Icon of Sin encounter but with a more challenging landscape to work against. Higher difficulties would add pre-placed monsters to increase difficulty.

Hope it works out! One other tip is to use at least 512x512 and downscale as the smaller sizes tend to produce worse results, I've found.

A bit of a cross-post, but I thought others might find this interesting. It's an AI tool that generates tileable material images. It can take in a reference image and/or prompt and has various configuration settings.

Use the keyword "seamless" to generate images that tile in all 4 directions.

"neon" and "cyberpunk" as qualifiers can net you some neat results as well.

I've noticed a lot of the output quality is, unsurprisingly, how well you can "zero in" which terms are well-represented inside the training set. You can feed the same image back into the tool multiple times to refine the results.

The best way I've found tends to follow the pattern:

1. Play around with terms to get something vaguely like what you want. Settle on a best candidate.
2. Pass the best candidate image back in to the tool and try more qualifiers to "hone in" on what you want.
3. Repeat until you are happy with the result.

This, like any tool, requires some learning through experience to use most effectively. But even so it's pretty neat for what it can do.

replicate.com/tommoore515/material_stable_diffusion

"Filler maps" in general is an insulting term. Its origins are from the music industry where hit songs would need to be accompanied by additional tracks in order to fill out an entire album, so low-quality, unoriginal or uninspired works were quickly put together to "fill in" those gaps, hence they were referred to as "filler." I consider the term always negative given what it means.

Pacing for a full map set is important, of course: "breather maps" have an clear place. Long maps need to be balanced by short maps, heavy action by quieter areas, etc. Without this the player will become overwhelmed.

Not treasure per se but I did find a real-life secret passage once.

I was in an old house and I swore I felt a draft behind a bookcase that was built into the wall. When I looked around the edges, I noticed that the frame didn't quite touch the wall. With some nudging and wiggling, the entire bookcase slid free!

Behind it was a large open space with musty wood beams and thick dust on the floor. No treasure, sadly. Just dust and spiders.

Ooh, I like this question. Neat idea.

For me, COMPBLUE / COMPSPAN / COMPTILE. Probably some COMPUTE1/COMP2 sprinkled in too.