Hello looters,

Here’s the latest news from the depths of the dungeon!

Summer’s over, and it’s time to double down to finally release Neverlooted Dungeon. In this devlog, I’m going to talk about the progress on polish and the latest level I’ve been working on, the last beta, which unfortunately went pretty badly, and other tasks that are less flashy but just as time-consuming and essential to make sure the release goes smoothly.

Temple Level Polish: Oops, I Did It Again

First off, let’s start with what went well this summer. I worked on the next levels, particularly level 5, codename “temple.” And… oops, I did it again…

You see, the level was already solid, and I was supposed to just do a quick polish pass. But, since part of my brain was stuck on the disappointing beta results (more on that below) and I needed time to let my thoughts on solutions simmer, I started working a bit more than planned on this level. Specifically, an idea that had been nagging at me for months: what if I added a little outdoor area to this level in a pretty cave… And, well, things kinda spiraled from there. I massively reworked the level with a ton of new content. If you visited a particular place in France recently, or played Dark Souls 3 or Elden Ring, you may recognize a few inspirations.

But rest assured, it wasn’t time wasted, quite the opposite. I genuinely think this level is now the best I’ve ever created. I find it gorgeous, varied, with a strong theme and an awesome atmosphere, fully explorable interiors and exteriors, tons of secrets, and fun puzzles. There are rarely seen features in such games, fully explorable roofs, breakable stained glass for alternative entrances, and tons of surprises. Honestly, I secretly hope some of you will consider this the best level you’ve ever played. No big deal. But maybe I’m getting a bit carried away, we’ll see what the feedback says at the next beta (more below).

To whet your appetite and help you wait a bit longer, here are a few screenshots and a modest little video montage to give you a taste of what awaits in this luminous temple… with a few dark corners.

Last Beta: The Cold Shower

For years, the game has been played, replayed, re-re-replayed, etc., by active community members whose help and feedback are incredibly valuable. But it had been a while since I’d had fresh players try the game in its current state. After weeks of prep work, including balancing, difficulty tweaks, bug fixes, etc., I was super confident about the beta and included a discreet little registration form at the end of the last devlog to accept about twenty applicants.

I was so confident about the beta that I might’ve even considered releasing the game in early access as it was. And then, the cold shower. Over 70% of players dropped out during or at the end of level 2! X_X Ouch ! At least, I’m glad it wasn’t the game release, that would have been a disaster.

Before all, I want to deeply thanks all playtesters for their participation, and a particular shootout to those who filled a lot of forms, and also to some who even took the time to record their playtest, that I slowly but surely watch. This is all very valuable feedback, so thanks again.

As I said, I had set up feedback forms for players to share their thoughts on the different levels to figure out what worked and what didn’t. The problem? Most of the players who dropped out didn’t fill out the form, especially for level 2, despite my many follow-ups. Maybe I didn’t clearly enough set expectations for feedback at the start of the tests. Luckily, I had set up (anonymized) analytics tools to track level events and understand what was going on. Thanks to the forms from those who did respond, some recording, and an in-depth analysis of game events, I tried to pinpoint the issues behind this massive dropout.

Looking at analytics, this player died way too much (red spheres is death) than it is expected for that level, showing a probable issue with difficulty.

Here you can see a player missing one secret, so if players miss too much checkpoints or valuables items that will increase difficulty even more.

So, it seems I made several mistakes. First off, since the players who were used to playing were Neverlooted Dungeon veterans, we completely missed the difficulty issue. It looks like the jump in difficulty between level 1 and level 2 was massive, leading to huge frustration and dropouts.

Next, I had recently reworked the attacks of one of the early-game monsters, the “zombie,” to make it run and jump at the player. But it was poorly balanced, and players were terrified, with the zombies being way too hard to kill.

Also, in level 2, some checkpoints were a bit hidden, and a lot of players missed several of them, meaning that on top of dying a lot, they had to replay a long stretch before restarting, which increased the frustration even more.

Finally, some feedback also complained about the atmosphere being too oppressive, the level feeling a bit too repetitive, and the objectives that could be a bit more clear.

In short, a pile-up of issues that created major frustration and a mass exodus. It’s a shame because the two next levels were all well-received by those who made it that far, mostly rated “extremely fun” on the forms.

Levels 1-4: Fixing the Beta’s Flaws

So, I spent the whole summer thinking and experimenting on this problematic level to better decide what actions to take. As it happens, for months, I and other testers had been thinking that this level would work better later in the game, so players would be more prepared and it would fit better narratively. My solution for the next beta is to move this level later and rework the levels, narrative, and balancing to account for it. I’m also working on fixing the raised issues, especially the repetitiveness.

Other issues were brought up for the levels, particularly level 3, which is very open and could leave players a bit lost. Since it’s now level 2, it was even more important to guide the player a bit more while still giving them a lot of freedom. For example, I started reworking the objective’s clarity and added a (super cool!) village map to make it easier to navigate. Look at this, so many shops to loot! Guaranteed* treasure!

Performance – Distant Chunk Optimization

One of the levels still had some framerate instability because certain areas were denser than others. And the temple work with its massive outdoor area led me to improve my custom scene-streaming system by adding partial data loading. This doubled performance on the temple level! Now I need to apply it to optimize the other levels even further.

Why It’s Taking So Long? Debug Example: Randomly Losing Interactivity

This polish is taking an eternity, and believe me, I’m the first to be sorry about it. I really want the game to come out so you can discover everything I’ve poured my heart into (and also so I can rest a bit -_-‘). But making a game of this scope, with so many features and content (15 to 20 hours of gameplay), takes a ton of time. Especially when you run into critical but random bugs.

For example, during testing, I regularly had testers report suddenly losing the ability to interact. But it was always weird, some interactions were sometimes partially possible. I spent days trying to figure out what was going on. In vain. Until one lucky day, a player sent me a save file where the bug happened consistently! After a deep dive, I tracked down the culprit! Turns out, some objects were occasionally clipping through the floor, especially objects created when others were destroyed (like bits of debris when you smash a crate). Well, those objects would fall through the ground and accelerate so much that their speed became near-infinite, causing Unity’s physics system to freak out due to floating-point imprecisions.

Since all my interactions are physics based, they started to be broken. For instance, here you can witness the physics raycast missing the book and the table in front of the camera and hitting the ground instead (the green ray with the white cross), making it impossible for the player to interact with them.

And yet, I had a system to prevent this, but some objects weren’t properly initialized. So, I spent two solid days building a robust system to make sure it doesn’t happen again. In these conditions, it’s always tough for me to announce a release date when I know I could stumble on issues like this, whose fixes can take anywhere from a few seconds to several days, like this one.

Register for the Next Beta

I’m working on the next beta, which will again be open to new players, hoping this time my changes will be enough for the vast majority to finish it while having a good time and discover that famous temple I’m pretty proud of.

So, if you want to sign up, if you’re ready to spend some time filling out long boring forms to tell me what you liked, and especially what you didn’t like, and even more so if you rage-quit the game and never want to play again, here’s the link: https://docs.google.com/forms/d/e/1FAIpQLSeeRiJdKlmvF-nMP23tNMx75PKZwsVAVIlBAjEFZdM-bVisFA/viewform

Roadmap

Time’s flying, the end of the year is approaching, and I have serious doubts the final release will happen before the end of 2025. But there’s “not that much” work left, so the release could still be close. It all depends on the feedback from the next beta, if everything goes well and we don’t need to make massive changes again, we’ll be almost there.

See you soon,

Arnaud