Hi there, and thanks for your patience! Making videos always seems to take longer than I think. (And I'm still finding my way towards better audio too.) Either way, hope you can get some enjoyment out of this video. :)
Elated to see your progress! I'm sure you're very busy right now but my offer from the first video still stands. I would love to hire you! I'm sure that you would find the complexity challenging but familiar. If the idea of bringing a project to life with more than a decade dedicated to the visual design of it is in any way appealing... i encourage you to reach out! You certainly seem to have a wonderful head on your shoulders as far as problem solving is concerned. All the best. -Digby
@@digbyfraze Thanks, glad you're enjoying it! I have a few projects of my own to take care of, so I'm not available for hire. Thanks for the offer though, and all the best for your endeavor too. :)
I thought you meant the turn blocks, but then I looked closer. You’re right! I never would’ve realized that the question blocks are literally just turn blocks with a moving ? on them if it wasn’t for your comment lol.
No way, you're lying.. *Checks video* WHAT. How have i never noticed that?! in all my years of playing that game and i never noticed what was in front of my face for decades. incredible. I mean it's only really visible for 2 frames but still! Now i can't Us-see it!
This is Big Boo, the legal counsel of Nintendo Co, LTD. I am writing to inform you that you have been served a cease and desist from using the Nintendo Co, LTD copyrighted term "Plumber". Nintendo Co, LTD owns all plumbers throughout the universe in perpetuity. In fact, don't even use your sink. Pipes belong to us as well.
And Bricks, Turtles, Dinosaurs, Mushrooms, Water, Air, Fish, Squid, Squid People, Swords, Gravity, Castles, Princesses, Bombs, And Planets, Will Take This All Away From Earth.
One thing that's more commonly used in Godot is actually composition, instead of directly going for inheritance. It's basically just the idea of having many several mini-scripts in Form of components, that all do their small needed tasks. (It's basically already what you are doing right now. You give your object a collision component, a Sprite component, etc.) The main reason for that, is so you can build your scenes modular and don't have to rely on the restrictions of the inheritated class, or always have to make new ones. So for an enemy for example, instead of having it have an entity inheritance or similar, you would make something like a "hurtMarioComponent" and a "bounceMarioComponent" (and any other kind of behavior) instead. Basically packing each single thing into a component. This doesn't obviously mean that you would need to use components for everything and never use inheritance, but more so dependent on what you already planned on doing with them, and if certain components could ever get used between certain classes.
Modern OOP does seem to focus more on traits that bring in different behaviors to compose an object and allow them to fit more roles than strictly single superclasses.
One thing worth mentioning for eventual code release: make sure you address copyrighted enemy names in code/don't release re-recreated sprites. Make enemy names something really generic (i.e. this isn't a bullet bill- it's a shooting bullet). Sprites should be fairly obvious as to the why, so maybe include in the readme expectations for the sprites the user provides themselves. You could just fire-and-forget and not have to worry about this, but it would help future sustainability for whomever ends up adopting/extending this project to not have to first strip copyrightable works out of DMCA fears.
I think they're aware enough, I'd hope they would release the basis as a standalone generic "classic platformer" package and have SMW stuff be a separate repo
The tilemap trick at 8:40 is incredible! And this series has been a great source of inspiration as a total novice to Godot and game development as a whole. My current goal is to make small projects to gain experience and work towards developing a feature-rich game with satisfying mechanics and controls, but it feels like an overwhelmingly daunting task to do from scratch for anything more complicated than pong or breakout; but watching you recreate Super Mario World's offers has been really encouraging! These videos are entertaining on their own, but I've also learned about nodes, paradigms, and problem-solving approaches that I haven't seen anywhere else. So thanks again for making this series, I can't wait for future installments and to see what else you get up to!
If you aren't already, I would encourage you to try some game jams! They have been helpful to me when it comes to getting started. I still have a long way to go, but ultimately, the big thing is knowing the tool you use better and how to go about using the tools (nodes in Godot's case) you have to create the feature you want. Ex. My first game jam, I wanted to add a way for the player to respawn close to where they died. I could have added checkpoints anywhere, but instead I checked to see if the player was on the ground for longer than x amount of time and if so, I set that as the respawn point. This worked great!.... until the player died on the moving platforms haha. That was the only thing I couldn't get to work and I ultimately ran out of time. But considering it was my first non-tutorial project, I think my solution turned out pretty solid. I think ideally, I would not allow the value to be set on the moving platform but at the time, I couldn't come up with that idea as I worked through various other things for the jam.
So many different skills on display here. The obvious ones are your game development and programming skills, but there's also the entertaining script, concise explanations, and engaging visuals to accompany the dense subject matter and make it consumable without much effort on the viewer's end. Really good stuff! Can't wait for more.
My dude, you are super underrated with this. I love the deep dive into SMW's physics and code and implementing those intricate and detailed concepts in the modern game engine. Super awesome, and I can't wait for the next episode!
About the statement at ~11:02 As someone who's been making a collision system in pygame which is similar to the one described, you probably wouldn't have to put the code to "push Mario away" in every solid. Rather, Mario himself could have code that pushes away whenever he touches any solid. This info probably won't affect your trajectory (and tbf my code is far from bug-free), but I thought it was worth putting out there.
5:14 I noticed the value you put for the timer got truncated by the inspector. If you'd like to use more precise values you can go to Editor Settings -> Inspector -> Default Float Step, and turn it down to a much lower value.
''Dayum, this video is so cool, fresh, positive and-'' ''When you make money, you make enemies''. ''DAYUM THAT'S COLD!'' Awesome vid! I'm learning just watching/listening to you! Keep it up! :D
16:55 it is CRAZY how unsatisfying it is to see mario bounce on multiple enemies in a row and not hear the bounce sound change pitch with each successive bounce
Wow! I've been recently saying "A Plumber for all Seasons is the best romhack I've ever seen". And I had this video reccomended separately and.... here you are! Bravo!
That's actually a fun & smart idea! I could imagine replacing all the sprites with custom Godot adjacent themes and likeness this project could easily become like a preview/project demo for the engine's capabilities and simplicity in features.
@@AkaNoBaka Ok cool idea... but I feel that's a lot of thought stemming from a pun I thought of in 2 seconds. But I'd certainly watch a series on that.
studying stuff that already exists is a good way to learn how to make stuff. i've already tried my hand at pong and flappy bird, which aren't the most exciting projects, but trying to add my own twists is hard. i've got a lot to learn
Amazing video as always! Here are my recommendations :) 4:18 - > -Tilemaps can have scenes as tiles, it'll probably make your scene tree much cleaner!- Dammit, you're too competent for my advice! (8:40) 4:30 - > -I think a KinematicBody2D would be better suited here? Not 100% sure but like 98%- Nevermind (6:23) Yeah I... don't really have any advice! You're straight up just good at what you do! Can't wait for the next one! I of course say all of this with *immense* respect for you and this project. I've just been using Godot for years and want to help you with some small tips here and there on how to best do things :) (if anyone is wondering why it's edited, it's because youtube's markdown is unlike every other markdown I've used)
Man, I've been having so many troubles making my own game in Godot and you somehow managed to solve a good majority of the problems I couldn't. This was super eye opening and I really appreciate you putting it out there!
Nerd here, I really loved your video!! I was patiently waiting for Part 2 and it was worth it!! I like the way you're referencing the original Game and explain everything. There is so much to learn ❤
man as you dive into the mechanics it's like... as complicated as you imagine any given part of the game was to develop, it's probably actually 5 times more complicated than you thought. it's a miracle they ever managed to make it.
There's a performance mod for Minecraft that makes chests and other block entities render like static blocks, this is great for stuff like beds and even doors because they don't have dynamic animations. But chests have an animation that can only be done on an entity, so the mod changes it back on the fly to keep the animation and the performance, just like Super Mario World's question blocks. The Minecraft mod is called Enhanced Block Entities.
heres a interesting, more performant way to tackle the blocks, theres a function in the tilemap node that converts in game coords into tile coords, which then allow you to basically check what tile of the tileset is at that tile coords of the tilemap. then you can just check the tile at the position above mario when hes jumping up (not when falling ofc), allowing you to simplify the block bounce fx to more of a visual fx rather than normal objects you have to place manually in the level basically the standard way for object to interact with tilemaps : 1. get object position 2. convert position to tile coords 3. check data of tile at tile coords 4. code goes here after
Hi there! I'm just barely starting out on my journey of making a 2d platformer and I've been absolutely adoring this series so far, it's super informative, relaxing to watch and for my little ADHD having brain it really manages to hold my attention, keep up the great work Wye!
Recently, I watched a video about Entity Component Systems (I'd link it if I could find it), and I feel that your examples of inheritance would be better suited for an ECS-based system. Essentially, in ECS, instead of inheriting functionality from parent classes/objects, functionality is instead added via child classes acting on their parent. An example of this being helpful is in the case of your QuestionBlock and TurnBlock. In SMW, P-Switches can turn TurnBlocks and UsedBlocks into Coins, but not QuestionBlocks. If you were to implement this using inheritance, you would probably have a child of Block called PSwitchableBlock, and try to have UsedBlock and TurnBlock inheit it... until you realize that Godot (GDScript technically) doesn't support multiple inheritance, and TurnBlock is already inheriting TurnBlock. Using ECS, you could have the QuestionBlock and TurnBlock scenes have a HittableBlock child node, which gives its parent (the block) the functionality to be hit. You can also have a PSwitchableBlock node as a child of UsedBlock and TurnBlock that allows them to be turned into coins when a P-Switch is activated. This avoids the multiple inheritance issue, because TurnBlock can have *both* the HittableBlock and PSwitchableBlock nodes. You mention composition near the end of the video (which I am presuming means each block has functionality stored as a member variable rather than being inherited), and that would also solve the multiple inheritance issue, but I feel that ECS fits better with Godot design philosophy and is easier to organize. This was a great video! I'm looking forward to how this project develops. (P.S. Your microphone is very bassy. I don't have much experience with audio stuff, but I believe that putting on a (probably low-intensity) high-pass filter on voice audio would improve audio quality a lot. That's my only complaint with audio, though.)
My buddy sent me this video as we were discussing Godot game dev. He made it sound like the series was all done, haha, so the withdrawal hit with full force as this video ended. I went and looked through the entire channel’s upload history and I found the Super Mario World stuff very delightful. Even the 15 year old videos are enjoyable and a little inspiring. It’s a very nice RU-vid channel. This new video series is especially great, though, and your voiceover is very effortlessly good, friendly and clear. I’m excited to see where things head.
I am so glad you're turning this into a series! Learning about the physics and game mechanics underlying Mario has been so interesting, and I really appreciate your approach to teaching Godot by first understanding the problem and then exploring the various ways it can be solved. Can't wait to see the next one!
The nice thing about composition over inheritance is that you can swap out the inner object during runtime, which can be very helpful for things like dependency injection. I'm curious to see how you'd use it here! Awesome video!
These videos are really well made. This is basically how I learn everything too. Take a thing that exists, reverse engineer it as accurately as possible, and then move onto something with different elements to teach me different features. Coding is so frustrating but so satisfying when it just works.
Duuuude, I've been using godot for years and I just realized you can inherit scenes from this one video! I've inherited scripts a lot but I didn't know you could do the same with scenes!!
Didn't think this video would be this good, it exceeded my expectations. I'm starting to learn Godot but I might go with something simpler than that... GG vid keep it up
As someone who's been learning Godot recently as well, it's fun to see/learn alongside all this progress Keep it up ! (Also, using packed scenes in a tilemap is insanely useful information I never even thought of before ! very neat)
Hey, love this series of video ! Highly entertaining and lots of learning value. Waiting to see what's next ! At 4:13 you say you can't put logic in a tile form a tilemap. If you don't already know that, you can write a script that looks for block tiles in your tilemap, erase them and create an instance of your block scene at that coordinates. Pretty neat !
Another awesome video in this series! A few thoughts: - For blocks that can be hit, I'm curious if an AnimatableBody2D would serve you better for stuff like interaction with other objects. The key change for this one is you can actually update the position and have it affect the position of other things in game. - Collision shapes can actually tell you what has hit them, but only when using move_and_collide which doesn't make sense in this game. I just wanted to point it out in case you end up needing move_and_collide in the future. - I will be curious about your experiments with composition for blocks/entities! Godot much prefers that method. I wonder if you end up with some sort of combo. Composition for base entity/enemy then some inheritance for types of koopa troopas for example. - For Mario's collision, I wonder if a capsule shape would work instead? This would allow you to slide around blocks just a bit and not get stuck on things? This is somewhat common in platformer to get some of this. Really excited to see more and be able to look at the player controller more if you get to post the code! Edit: cape physics..... that will be interesting to see if you decide to try to create it!
I have an ambitious coding project using what was going to involve snes register values during SMW gameplay as live value inputs for a custom program that drives a wholly unrelated piece of tech. But this project of yours seems like it will blow the doors wide open for more possibilities! It’s exciting to see your progress!!
Excellent, informative, entertaining and chill. Very nice video overall about the pinnacle of platformers, well done! You briefly mentioned SPAWNING/DESPAWNING and this is one aspect that I find fascinating and there's a huge variety in how games from that time period handle that. If memory serves me well, SMW does have a bit more complex system, whereas many other games simply make the "entities" disappear when exiting the screen. At this time it was quite normal that you couldn't freely move back around the level, so that was basically the reason why a complex system was not needed. A common strategy in some games was to get the enemy outside of the screen instead of killing it. In SMW you need to at least save some of the information about what has been interacted with and how entities are still "alive" and what they are doing when outside of the screen. This would be interesting to hear more about - if you find it interesting too ;) Cheers and all the best! Looking forward to the cape physics but also the different Yoshis and the really convoluted special stages with lots of secrets and hidden exits - if you have the patience to go that far.
I have decided I will watch this series and apparently there are only two videos for now so I’m subbing and hitting the bell. Please don’t end the series.
I think I've learned more from your 2 videos than I have any other tutorial. I really want to make games that emulate the Snes, but I always end up fighting godot instead of working along side it. Thank you so much, very inspirational.
I know this video is a little bit old now, but for anyone looking for an answer. You actually can detect collisions without an area 2d because the built in function 'move_and_slide()' creates a list of everything it hit every frame including the details of the things it collided with
Also for testing if it is only Mario touching the object, instead of using collision layers it would generally be better to simply and Mario to a group like 'mario' and change the area entered are to a body entered area and do a quick if body.is_in_group("Mario"): You code
I'm making a platformer in Godot and these videos are helping me a lot in which things should i consider when making these types of games. its a cool job you're doing here 👍
Love the video! one thing I'd like if i could tell what's your recreation what what's the original game. but every thing else is incredible can't wait for part 3!
I'm really glad you're trying to follow the same transpiled logic as the original game, I really wanna do something similar and this series makes me happy because it's literally what I would imagine myself doing.
This is amazing as I'm just now learning Godot myself. The only thing a total noob like me could probably tell you (if you don't already know) when you are setting up collision for your tiles in the tilemap is to create the physics layer, select all the tiles you want to have collision, then in the select tab under your physics layer hit "F" on the image. It will create the default square shape on all the tiles so you aren't having to create all 4 points every time and then you can just select one tile at a time to edit it the way you would like.
In Godot, Rigidbody2Ds can detect collsions, so I tend to use those for a lot of this kind of thing. Though they're also supposed to move, so it can be awkward with solid objects, but it's solvable.
I've also been doing some Mario prototyping myself in Godot. Upon watching your video, I realize I should probably group game entities a lot more than I currently am. So thanks for your insight, wye!
Swear you have the voice of "@MattKC" on YT. Ever so slight difference, but it'd be VERY hard to distinguish. Excellent breakdown (and love the mention of RetroGameMechanics Explained. Love his channel also!
I really think you should make templating a secondary goal. You could potentially help thousands of people learn to understand the intricacies of platformers. Your mario world engine could be a great learning tool.
This video seriously gave me inspiration to try to learn Godot again! Thank you for the high quality videos you produce, i seriously can't wait for the next video or the next project!
I know that if you did you'd basically be begging for the nintendo ninjas to start busting down your door, but it would be pretty awesome if this eventually culminated in a fan-made mario maker port lol
Hell yeah, king wye is cooking again! Many of the things you talked about here sound awfully familiar to things I'm doing in my job. We also handle "level collisions" and "entity collisions" separately. (And yeah, we also call them "entities"). If you ever get further into implementing enemies, I'm really curious to see how far you'll get using just inheritance alone. In my experience, inheritance can be a very useful tool for a limited range of use cases, but you'll eventually run into issues once things diversify too much. I had this problem when creating entities in our prototype using inheritance. Let's just say it really wasn't a long-term solution. For our acual game, we've created our own implementation of an ECS (entity component system), which solves many of those issues. Basically, whereas inheritance tries to describe entities based on what they are, an ECS more so tries to describe them based on what they can do. An ECS an inheritance can be combined, though the latter will usually be used quite sparingly in that case. I'd say it only makes sense between entities that are very, very similar. I don't know if Godot actually features any built-in ECS implementation (even Unity and Unreal Engine only added support for ones in recent years), but I feel it would work wonders for an SMW recreation. In fact, I'm like 70% sure the original SMW featured something similar to an ECS (based on some code I saw in a disassembly, and also based on the fact you can customize original sprites so easily via the Tweaker application). So hey, since you probably want to gain some general game dev experience along the way as well, it's probably worth looking into an ECS. There's even some open-source implementations around, like EnTT or flecs.
im really enjoying this series so far ! as someone who's got no experience with godot yet and is looking into making a 2d mario fangame, it certainly helps with my confidence to see someone else tackle a similar goal. i'll probably reference your code a lot when you release it haha
This is definitely some good stuff right here dude. Once you do a full on Godot remake of SMW, can you also try with the SNES Donkey Kong Country Trilogy maybe?
are you sure you are just learning all that stuff? cause you are doing everything perfectly/subjectively optimal. as an indie dev myself I can say that this project is better put togheter than 80% of indies on the market. great video and project
Little suggestion/musing/whatever- while using inheritance for block behaviors is good, I think it would be a better idea for the Mario controller to be the one checking for interactions with blocks and then 'asking' the block scripts what should happen. You can also replace the area2d interaction points with raycasts, but it doesn't matter too much. While the main reason to do this is because it reduces the amount of collision checks being done and thus is a performance benefit, I find it significantly easier to debug in the long run also. Plus, without having too much familiarity with SMW source code, I reckon it would be more accurate to the original - as you said, in the original game the 'blocks' are not sprites but background tiles.
