Literal recreation

I took a day to do whatever i wanted, and found myself reworking this screen. The difference is large enough to show and tell / ask for critique.

It seems to me the dither on the coarse rocks of the castle foundation is done better justice now that not everything is made out of dither. I also helps that the fog looks more like fog, instead of static. I'm not satisfied just yet, though.

Hot dang, that looks awesome. What a difference a year and a half makes.

Definitely looks better now that it's not swimming in noise, and the stingier use of purple makes it much more eyecatching where it is used, in the best way possible. I like the emphasis of color, contrast and outline too, especially in the darker areas.

Some of the changes between blue/purple and green/yellow seem a bit arbitrary though. Like attribute glitches that aren't quite hidden. But you have to be looking to notice them, so it's not that big a deal.

Thanks! It feels good that that time wasn't spent in square one.

At first i had thought this to be some sort of splash screen for an intro or even a castlevania-style ending scene. But with a flat enough level segment, it could really serve as a background. Like so:

Some areas (most notably the fog creep between the castle and the lesser mountain) will checkerboard-shimmer this way. How much shimmer is acceptable?

Oh wow, that looks amazing

Amazing work! That's all I have to say..

That parallax intro is really pretty.

Wow! You create wonders.

This is lovely, both as a background and as an intro - I think the parallax scrolling works for both.

Super duper incredible stuff! Really want to see this further developed into having sprites doing something.

Aww you guys!



For the purpose of scenery - If used as a level, i was thinking of maybe having a celestial body rise or descend really slowly, based on time passed since level init. It would look slightly different from one playthrough to another based on how quick the player is. Used as an establishing scene shot/intro, it could simply be timed along with the rest. A place i work and live at each summer has a geography where fog filling the valley often coincides with sunrise and sundown.

I initially thought of lightning, but fog and lightening in the same vincinity at the same time is a rare phenomenon due to different meterological preconditions - But that didn't stop the sorcery from happening in the mad forest of Dracula's curse, though

jesus christ, let's make this game when we're done with the compo! i'll even put up with castlevania-style mechanics (except the stairs. i can't do that in good conscience)

If you can handle me trying to convince you at least once how we might approach stairs in a better way than castlevania did, do we have a deal?

in all seriousness - i will try to emulate castlevania movement exactly if you want.

that said, we really should improve the stairs though!

Rondo of Blood does Castlevania stairs perfectly.

But honestly, if you don't appreciate Castlevania-style mechanics, please don't make a game in that style. Make something that you enjoy yourself and are able to get the best out of.

Yeah, jokes aside (we had a previous discussion over mail on how a certain object in our current compo project should work where castlevania got mentioned a lot), i think it is unlikely we'd come up with something exactly like the castlevania classic mechanics, partly because the process would take us in directions, and partly because it ought to bring at least something new to the table and get an identity of its own. There's already a possible deviation in that mockup in the form of 8px platform elevation differences, for example. And that scene alone makes it impossible having bg based status bar. I'm just very keen on proving there's room for a game in the good old actionvania fashion.

Will look more closely at those rondo of blood stairs. The fact that you can jump on and off makes me believe there's two objects (cannot validate atm) or at least two cases within that object: the usual stairs and one that leads up/down through the floor between neighboring screens, because iirc you can't jump off stairs while moving through to the screen down under?

Had another anti-stressant doodle session.

While this mockup contains two unique columns of distant background on each side of the tower, sideways scrolling can only show one side at a time. Which means the same two columns (instead of 4) could be used on both sides without the player realizing it unless actually studied.

For it to fake parallax it then needs to "scroll" ~10 unique tiles (times 8 prerendered scroll positions), and 2 more unique tiles times 2 , so 84 tiles, i think... which could actually fit inside the bg-chr page and still have a lot of tiles to spare with this set or something like it loaded in. That's the brute-force approach, at least.

Normally, a tower would scroll the other direction, but i think segmenting it like this somewhere could have some gameplay value, too. When/if vertical scrolling is needed, the screen would just omit the outside parts to keep things simple.

Anyway, if you have any suggestions, let me know!

Looks great! I love the texture of the outer walls!

It took me a minute to realize there was a window on the left wall, and that's why there are lighting effects coming from that side. I'm not sure I would have noticed that in-game (I only notice because I'm looking at a static screenshot). I'm also not sure about the perspective with the two windows; I think it's just their placement, and the fact that the window on the right is the same height as the one on the left. It might help to make the one on the right shorter, as it's farther away. I do like the intent, though!

The quality of the drawing is incredible, but there is definitely something wrong with the left window's geometry.

To me it looks like it was purposely designed to try to force your brain to view the scene as a curved tower (back wall being concave) instead of a flat scene. The angle of that window is different from the other because the wall is a different angle at that point.

But then the ray of light should also have a different curve, which is not the case here. Also it would seem the wall is way more curved than what would be consistent with the border's position if the tower was cylindrical.

Thanks everybody, this is great input!

The tower was/is indeed meant to have a circular wall, but there wasn't much to support that intention. Here's some few changes i made today which i think conveys that better. Just some clean-ups to do, before i think i'm satisfied.

Yeah, in a realistic scene, rays of light would be cast differently from each window. I tried that first, but kinda failed, so i went with this more stylized/unrealistic setting. I might attempt it again, but there's a couple of problems:

-at this gritty grainy style of detail, rays in several directions would eat up the rest of the tilemap.
-the alternative is coming up with simpler rays of light (another sort of stylization)
-but even so, some angles just doesn't look that good at nes resolution.
-which can be somewhat hidden with a bit of noise, in other words back to a grainy style.

So, that's the carousel of compromises i got a bit tired of eventually, and just went with diagonal lines as my base. Not saying it can't be done, of course... I'm just not sure it fits my preference.

Seems a very nice dungeon!
Maybe you could add some skeletons, be it complete or in parts.
If it really is a jail, of course.
Maybe adding some sprites to create a smoke or fog effect on the animation would make the ambient even darker.
Keep the good work!

It is a very interesting scenario.

You could play with the background, creating clouds that pass slowly from left to right; and occasionally make a change palette to blue tones of light pass black and disappear, leaving just the white color outlining platforms.

It would be very nice.

Diskover: Thanks! "interesting" is one of my keywords for these scenes, so i'm happy that was the word you chose, deliberately or not. More explicitly, the goal i'm excercising towards is to tell something about the world in as many scenes as possible.


Fischer: You're on the right track. A prison is a likely side-use of a tower. I want the imagined player to ask "a prison - for what?". Note the unnatural size of those collars. That said, it's a detail that might be moved to some other scene in the end, who knows.

Right now i'm more or less free-form doodling pieces as a way to relax and focus, and stritching those together into a string of levels that make geographical and narrative sense would come later.

I'm kind of fantasizing every level would present a high level of uniqueness, but given the size of an average cartridge, it would be practical to reuse some of the settings at least a bit. I think this could serve both as a tower standing on the abutments of a bridge, and a part of the castle itself. A few palette changes should help.

Both: On sprite addons, i was thinking about maybe slowly animating the rays by moving diagonal lines slowly (in equally diagonal paths) back and forth in a sinoid cycle, if that description makes sense.

I like it, but is that the darkest blue the NES has? I think the glowing light would look better with 2 shades of blue.

Heh, oops no. I guess i must've grown used to the blue emphasis bit being set.

This attachment has all emphasis bits cleared and i changed most of the ray highlights to blue. Nothing else is changed so far. Might be a keeper. Thanks for the suggestions!

Think about it, you can look good

Maybe if you exchange the position of the left window (the one which is visible, not the one which is suggested) and the chain it would look better, because it really look like the left window is at 45° angle from the viewer, which is strange since it's just next to the central window which is parallel to the viewer.

(That's what I tried to say before but I probably wasn't clear...)

Ah, thanks! In that regard i may have worsened the look between the revisions. I'll look into it.

Just wanted to say, the revisions look much better! Is this part of the level scrollable? If so, what happens to the windows and the curved bricks when you scroll?

EDIT: Woops, missed the part where you said this was more "doodling".

I agree -- the window angle suggests a stronger curve of the outer wall than would make sense. Moving it left would make the whole thing work better.

I find it slightly easier to show the scrolling concept than try to elaborate it in words, but i'll attempt both.

But basically this tower would force-stop a horizontal scroll quite early in both directions. There would be no vertical scroll in this particular case (even if it is a vertical climb. It means each screen is presented instantaneously just like in a non-scrolling game. I think that's a nice thing to diversify the experience the camera provides, as opposed to NES metroid for example where a vertical shaft will always have vertical scroll and so on - no surprises there- CV3, by comparison, has a number of perceived camera modes: upwards ratchet, downwards ratchet, auto-scrolling, coarse scrolling.. helps diversify the various towers. Scrolling/non-scrolling in the general direction the player is heading would be the simplest form of diversification and i think that's a good start to present different challenges tied to either condition).

Besides, this cross-section:ed concave form doesn't lend itself to vertical scrolling, because it's concave in two dimensions. It'd have to one-dimensional for it to work together OK with scrolling along the length of the tower.

In a real game situation, i'd like to think that the camera would "want" to be at the center but you would be able to drag it out to the side/s by moving close to the wall, sort of like pulling away a magnet.

Those are my thoughts at least.

This mock-upped visualization contains a few demonstrations:
-Tile selection-based parallax (eats a considerable portion of the chr table. Animates tiles by pointing to left/right shifted tiles that are pre-rendered and kept in the table, Pushes in the opposite direction of the scroll to maintain a fixed position.)
-Some sort of foreground object in the form of a backlighted chain (i don't feel too sure if it's look is final - it might need a bit of polish and maybe another shade of blue/bluish).
-Trying out some semi-independent movement of the same.

How does it all fare?

Oh, and i reworked the windows so the left one isn't at such an extreme angle. I saved the 45-degree one for reuse in some similar screen so the level design can play around a bit with components like windows an chains and not always keep them identical, screen for screen.





Maybe doodling is the wrong word, idk - I almost always have a goal of some sort when drawing, but basically i've found this setting to be unusually relaxing and recreational to draw. At times it's almost like being 9 years old and draw faux stages to some favourite game.

Geeze, you never stop amazing me. It makes me sad that I could never do something that cool with NES graphics, and I used to be a pretty decent artist, too. :\

Sumez: Well i'm pretty sure you can make stuff i can't (like that avatar you have), so maybe that's even? Btw i tried to search for graphics you had done, but couldn't find any OP in this subforum. Do you have a blog or something?

Diskover: I'm sorry i forgot to reply - thanks for the demonstration! I'll certainly consider it. For one thing, it's a good effect to simulate clouds of the low-hanging, dense type floating by; affecting the overall ambience. It might also work as a nice (maybe somewhat quicker) fade-in, which would be a bit more subtle, i think. Btw, what tool do you use to mockup those palette shifts?

I made that show with Photoshop. It was just to give you an idea. In a real NES would not be the same, but very similar.

The pallet change is made from code. It is not complicated.
If you want you can pass me the CHR and the NAM of the screen and I make a demonstration to see how it would look like.

That parallax scrolling looks great but it looks like it will "eat" a lot of sprites. If the background was just made of horizontal lines, it would need no sprites, but then you can use sprites only for example for the trees, and leave the sky only with horizontal lines.

Having the left wall not purely vertical-shaped also requires more sprites, so does that chain. So as long as all those objects (trees, chain, non-perfectly-aligned-left-wall) are not on the same scanline, it's fine to have them, so that the effect overall just takes one sprite per line. Incredibly clever

PS: Since that chain is a sprite, you could also made it swing one or two pixels, just because.

I was hoping to have a technical discussion on that matter. The way i planned it, it would not eat any sprites at all, because it's not "true" parallax. For the lack of a better term, i'd call it "tile select animation" or maybe "tile select parallax".

What you have there (or rather a mockup of) is a background tile animation. Basically, something like a third of the the whole chr file used for this scene is filled with left/right shifted variations of the same few treeline tiles. That's what i meant by "brute-forcing it".

The map would have a property that made certain areas' tile ID:s an index, and then while scrolling, they would shift to a another tile to make it seem like the far-field background is still, when in reality, it is not: It's just animated at the same pace as the scrolling to counteract it.

That solves the problem with the windows' indentation, too.

That would never fly back in the day, but since it's no financial problem to have a 512kB cart these days, one could afford the extravagance.

It would also only work because you cannot scroll longer than two tiles away from the tower, and the tower itself is hiding the fact that the far-field background has a very short report; 2 tiles wide (in a virtual sense: 3 tiles, if you take the full contents of the tile select animation into consideration).

It never occured to me the sky gradient could be sprites, but regrettably, i don't think that's an option here because we'd need a sprite based HUD. On the bright side, tile-select fixating the gradient wouldn't require as many tile variations as the forest: Counting, i see 2 of them needs 2 tile variations to be kept still, and the 2 others need 4 tiles. so, 12 tiles in total is the price for a spriteless faux parallax sky gradient in this case. The price for the spriteless forest is more like ~80 tiles

And just to be clear from the start: There's no continous chr ram loading, since everything is stored in chr ram at the same time. It means i can't add too much extra/unique details into this scene beyond what i've already done. I'm a bit confined to just play with their placement from room to room.

--

Oh yeah, i was thinking the chain could perhaps dangle a bit in a pendulum like fashion; just maybe, if it isn't too much attention grabbing. It might be enough that it masks the player/an enemy and moves while scrolling. But if it did dangle, it would make sense for it in that case to left-right-move a bit link for link in inertia, if it can be pulled off along with all the rest of the stuff a real game would need to do.


Same here. In PS, i would open a copy of the image, ensure the colour mode was indexed, then use the colour replacement dialogue. Then copy them to different layers in a new psd file, rince and repeat three times, and lastly animate the visibility of the layers using the timeline window. I'm asking because i have a feeling that's maybe not the best way to do it.

Well, having the side trees rendered multiple times and changing the tiles on the fly when scrolling makes sense, this avoids eating sprites. However I do not think having that dithering on the sky gradiant scroll makes sense, especially after applying NTSC filter. I don't think it should eat that many tiles, why do you mention a 512kB cart ?

I do not see how a sprite-based HUD is conflicting with it being a sprite, as long as it's never on the same scanlines.

Leaving the sky-gradiant dithering as it, but scrolling with the main background might work but it might be obvious it scrolls at the frontground's pace and look weird. So I really think the optimal would be to stick with horizontal lines. Wasting tiles to represent a scrolling dither pattern really doesn't make much sense.

Your work is amazing, I could never draw such beautiful graphics.

You do not have a CHR? Everything you've done with Photoshop?

Well, then I'll try to recreate it. . .

The truth is that with the few colors that have the NES palette, this effect is difficult to do well.

That's the thing. Ideally i'd have the sprite HUD in the upper left corner of the overscan safe zone, so i don't want to cause too many sprite overflow conflicts there. The chain adds 1 per line, the player going up the stairs adds ~2...

Since they would appear as still, i didn't think it'd be a problem (ie if it was moving, we'd see dancing artifact colours)? But i should probably check how it looks on hardware sometime soon and verify this as good as i can (both my units are PAL).

This got me confused (i hope i'm not missing something crucial?) I wonder how it can be obvious it scrolls at the backgrounds' pace? The tile animation should be able to completely prohibit the gradient from appearing as scrolling (even if it techically is).

An analogy might be in place: If you drive a car on a big treadmill, and the treads of the mill scroll backwards, and you drive forwards at the exact same speed, you remain in the same place in the eyes of anyone observing. That's what the tile animation should be able to do, towards the result as shown in the mock-up.


If the tree lines are using 80 tiles alone* in this tower section of a level, that's a trickle. In my previous mockup (the one with the castle and the low, flat stage), that scene alone is using up all 4kB:s, so it's a somewhat bigger trickle. Together (if they were strung together), they do not really make up a complete level of average actionvania length. So imagine many trickles adding up to a stream that makes up a level, and then add together a bunch of levels and you have a river wide of graphics and level data. Getting detailed and custom like this for a complete game would require those 512kB:s, and using 512kB prg-rom was mostly not in the budget of most historical NES games (128-256kB seems about average?).

*actually, it can maybe be made with some fewer than 80 tiles and still look roughly the same, because i just might be able to reuse some same tiles in different attribute cells. That's for another day to experiment with.

Thanks, i really appreciate that you've taken the time to discuss this.


No, that's not it. Everything is done in NESST, except the rendered camera movement and sprite layering of the chain, which is done in PS, in this case. Sometimes when i want to get fancy, i use AfterEffects instead, especially for sprite animation reviews.

I could never draw that thing as seen from scratch using PS - it's too far away from how the NES graphics are laid out, + NESST helps me keep track of my tile budget.

Sorry, i sometimes feel hesitant sharing the chr:s of works in progress because i don't want too many "incomplete" versions floating around. It's just me being a bit sensitive while in the process.

It doesn't shimmer unless you move, yes, but that's not the only problem. The NTSC rendering has a repeating colour-error emphasis every 3rd pixel of a scanline, and then shifts diagonally one pixel each line. If you make a checkerboard dither, there will be a repeating diagonal wave of emphasis 6 pixels wide across that dither. (Whether this is visually acceptable is a judgement call, though.)


I don't think anything is obvious but...

My initial assumption was that a tower level would have vertical scrolling (and thus vertical parallax). So I was surprised when you showed with that GIF that you meant to shift the parallax background only horizontally. Is this screen meant to be static vertically then? Your estimates for tile permuations sounded like you only meant for 1 dimension of parallax, not 2. (Incidentally, using sprites could make both feasible at once?)

Would i be able to properly view how this looks on an AVS on a HDTV in Europe? I've been thinking about treating myself with one (to be able to play N.American releases properly, but also to be able to test things like these out).

But yeah, these two mockups are so-far pretty heavy on dither in places I think it just might be the style i want, but i may want to revise it if i find it godawful on HW.


Yes

I figured there's some merit to it being the other way around (not as a general rule for the game, but as a mode among others you can have in some level segment to vary the overall experience).

Vertical progression + vertical scrolling, and
vertical progression + vertically static screens

will provide very different opportunities to present platforming challenges, enemy behavior and so on. Think about Battle Kid 1/2 and how much the gameplay relies on a scene/screen being set up all at once for its platforming puzzles to work. It's also different from the church tower in Warakiya village in CV3, where the pit of death is constantly changing depending on your elevation because of the ratchet scroll, but in my mockup, pits of death are fixed-point.

Another reason i wanted this segment to scroll horizontally is because i'm thinking of ways the player could interact with the outside (i don't want to spoil how at this point).

<Digression>
Another study case i find interesting is the clock tower of untimely death in CV3: You go up, and after defeating Grant DaNasty's monster form, you go back down. The scrolling mode is ratchet scrolling which herds the player character in the right direction. The part i find so interesting is that they change the direction of the ratchet, so first when you scale the tower, if you fall down, you fall to your death. But on the way down, you can take plenty of shortcuts by simply falling. Besides the obvious point that it lets you get back to the door, i think they made it this way so that the downhill travel would intentionally take a lot shorter time and feel different (which makes a lot of sense since the level is reused). The fact that Grant is faster, jumps longer, can control his trajectory while jumping/falling, and climb walls also plays part in the shortcut taking/quick backtracking, all while keeping it fresh.</Digression>

Of course, you can do some timing-based platforming puzzles in a vertically scrolling tower, too. Above mentioned clock tower in the digression, for example, has a few instances where the spawning of a medusa head is triggered based on player elevation (you set it off by a jump that forces the scroll upwards). The puzzle here is to sync the enemies' appearance nicely with the cycle of constantly spawning medusa heads to make life a bit easier while maneuvering through the building.


This is also correct, i intended it for one dimension. Plus the way i'm using different attributes in the forest horizon should prohibit vertical parallax... I'm having a hard time wrapping my head around how sprites would help with 2-minensional parallax, though?

I don't think so, since one of the main points of these new consoles is to generate video that's more friendly to modern TVs. If they even have composite video out, it's probably "cleaner" than the original, without the hacks that made the old video signal so peculiar.

Thanks, that makes a lot of sense. So what i should be getting is an old NTSC unit shipped here.

In theory, an HDMI famiclone could emulate NTSC modulation artifact emulation in the same way that Nestopia does. But this video walkthrough of the AVS's video options doesn't show such a filter.

But then you can just as well use Nestopia (or other emulators that support NTSC filtering) for testing and spend no money at all, if you're only interested in the artifacts because you're an artist.

I personally would love if these new consoles supported accurate NTSC simulation, I simply love those artifacts.

Someone else might correct me if I'm wrong but I think the artifacts come entirely from the NES' signal generator, it's not really the TV. (Though the TV obviously makes its own impact on things.)

Lots of NES emulators have an NTSC simulator in them that do a very adequate job, IMO. I usually use this build of Blargg's filter demo to preview this kind of thing when I'm working with images instead of emulators. (A lot of emulators that have it are just using Blargg's implementation anyway.)

BTW the diagonal "volumetric lighting" spilling from the windows will be heavily affected by the artifacts, and will definitely shimmer as they move, but by coincidence the artifacts are at the same diagonal angle, so you might actually like this effect.


Of course, but I wasn't stating a preference. I just missed the point of reference to suggest one or the other.


Most of the strip could just be one colour 8x8 tiles, and then sprites for the detailed parts at the horizon and top of the screen. You'd only need one copy of the tiles because the sprites could move up/down/left/right as needed, and the coarse background tiles would not need to be updated frequently (could probably get away with changing only one tile at a time as the horizon moves).

The left/right motion would need some way to cut off the sprites behind the background, but there's multiple ways to accomplish that. (e.g. placing the sprites behind an opaque background, or if you can spare a third sprite on the scanline you can use a blocking background-priority sprite to hide it, or you could use horizontal tile permutations for the cutoff and only shift the sprites vertically.)

Aha, so the background would only be lines of colour, and the sprites would provide the detail.
I might change this, but right now, the wall is completely opaque (the bg colour is the bright sky one, and the black is shared by three subpalettes), so that's a start.

I'll keep this idea in the options drawer for posterity*, thanks!


*(it'll take some time before we can actually work on this properly because we already have one game to finish).

Another (simpler, less impressive) option is to simply center the camera so nothing outside is shown when scrolling vertically.

If blarggs' tool is accurately showing these artifacts, i don't think the still image looks that bad? I'm a bit pleased it seems to smooth out the gradient a bit.




One last thing is in the way of making this scene work as a vertically scrolling room, and it's the 2D concaveness of the cross-section perspective (type B). Type A solves this but makes everything look oddly tilted. Two other options are stretching the vertical concaveness out over 2 screens height instead of one, or making the shape of the tower concave in only one dimension, which weakens the effect by some, but works endlessly.

That might actually be neat! Unlike the shimmer you can see (but which i've sort of accepted) in the first outdoors mockup in the lower mist and mountain base.

Well, like I mentioned, those diagonal light shapes are strongly affected. You can see a lot of diagonal groups where it's discoloured, but it complements the texture nicely there, esp. because the angle of the artifacts is the same as that diagonal direction anyway.

The regular dither patterns above the flat stones, though, you can see how the varying emphasis the the regular shapes there produces strange grouping patterns. This will shimmer and stand out very strongly as soon as you start to scroll as well. (Especially note how about just below the middle of the image, there's lines of 4 pixels that are emphasized. That will change with scrolling, dimming and brightening every 3 pixels.)

If you want to try it out without using an emulator, you could make 3 images with the tool, each with 1 more column of pixels on it to adust the phase of the artifacts. (Note: the phase changes only when you scroll-- when standing in one place it will not shimmer, so it's going to be driven by player movement, rather than being a continual animating thing like this GIF.)


It's a subtle problem, though, not necessarily something you have to worry about. Unless you cover big parts of the screen with blocks of checkerboard, I think it's reasonable to just ignore, even.


Honestly, I think what we're looking at here is too subtle to communicate a curved back wall, and the prominent flat foreground blocks strongly contradict it anyway. The parallaxing hanging chain is a nice touch (but doesn't really have to do with the curve). The tilted window feels weird to me without already trying hard to imagine that it's curved, and similarly the curved dark background bricks feel more like random variation / uneven stonework than a suggestion of curve.

I like the attempt, but the effect isn't strong enough to unambiguously say "curved tower" to my brain. At the same time, though, I probably wouldn't want to spend as many extra tiles as it would take to imply that more strongly anyway-- as a square tower it already looks great. (...and I probably wouldn't even think twice anyway if there was an external long shot view showing that it was really curved from the outside. A square inside would be an acceptable fiction in this context?)

I really like that constant shimmer in the light rays in your demonstration; it feels alive. It makes me wish i could achieve that constantly without shaking the structure. It has fantasia's night on bald mountain interpretation all over it (which left a huge impression on me as a kid). Then again, it might be too attention grabbing if it did cycle all the time.

I guess the misty highlights along the floor could be tamed just like the sky gradient, but again at the expense of hogging a few more tiles. Or i could come up with a more stylized option for about the same number of tiles.

I see. It might be that my intention and preknowledge has got me fooled. I'll have to think about that. I'm pretty sure i can do a better job, but i think i'll let this scene rest for a while, and see what comes naturally later on.

This is all very helpful, i really appreciate it.

Normally, the NES PPU skips one dot before every second video field in order to make the dot crawl pattern appear stable. But it doesn't skip the dot if rendering is turned off. So to get that shimmering on an NES, you could leave rendering turned off for the first scanline for two consecutive frames out of every eight, ten. This may be tricky to time and may cause (invisible) flicker artifacts in the top overscan, I'll admit.

I bet if you're scrolling up, mostly, the inverse of type A could work because your eyes are focused more on the top of the screen, where the background would appear to be in the correct perspective.