That's the thing that gets me. It seems like a lot of the time, hardware design is more about guessing what developers will like, and then kind of crossing your fingers. I wonder what the reaction was from the hardware team when they saw R-Type for the first time.
I think that you are forgretting that Hudson was primarily a software-development company, and that they designed the PCE to overcome the limitations that they hit when developing games for the NES. They had the passion and guts to completely design their own CPU/PSG/VDC/VCE chips and build an integrated system that is (IMHO) far more elegant than anything else in the 4th-generation (elegant doesn't mean more-powerful).
That is the opposite of Sega, who were primarily a hardware company (producing custom-board arcade games), and where the Megadrive was designed by a hardware guy without ever talking to the software guys (as he admitted in an interview many years later on). Sega were used to buying as many off-the-shelf parts as they could, and then sticking them together to meet a price-point.
Nintendo were somewhere in the middle, but with folks like Miyamoto at the management level, they focused on software capabilities as much as hardware production costs.
If you think about it, almost every bit of technology and hardware design in the PC Engine is an evolution of stuff already out there. Nothing the PCE did was really new in the landscape of console hardware. It was the same thing that was already out there, just MORE of it.
Well, apart from the fact that the "evolution" that you are describing applies to every generation of consumer gaming technology, you are also ignoring the fact that the NEC's vision for the PCE was to sell the new CD-ROM technology, which was revolutionary at the time (and damned expensive).
It is hard to know what their design calculus was. I suspect they were going for a combination of cheap and good enough, ...
You have to understand that every consumer-product design is governed by considerations of "cheap enough and good enough".
You also have to look at the massive technological leaps that were being made in the 1980s that effected the calculations of what manufacters could produce, and at what price-points.
RAM size and speed and price were critical design factors. RAM speed itself was pretty darned slow in the early 1980s, getting faster and faster as the decade progressed.
The machines that were designed in the early 1980s (like the NES, Colecovision, etc), were designed to use the 64Kbit (8KByte) RAM chips that were in production at the time.
The PC Engine was designed at the time that 256Kbit (32KByte) RAM chips were becoming affordable for manufacturers.
The SNES, being designed a few years later on, was able to use the newer 1Mbit (128KByte) RAM chip for its main memory.
Note that all of those chips are 8bit (1byte) wide, which is why we get the 64KByte size of video memory on all of the 16-bit generation machines, being 2 32KBytes chips.
I suspect they were going for a combination of cheap and good enough, and saw themselves very much as the natural successor to the NES. A second background layer might not even have been a consideration, since nobody else was doing that in home hardware at the time.
I think the more important question is why didn't the OG PCE just have the VDC capabilities of the SGX.
I'd take more layers over more colors.
If you look at the VDC's design, I believe that you'll find that it was an issue of the RAM speed at the time when the PCE was designed, and RAM costs.
Given the same speed of RAM, the 16-bit generation VDCs could process twice as much data as the 8-bit generation VDCs.
Thus the NES's 4-color per sprite/tile (2-bits) in its 3rd generation hardware becomes the PCE's 16-color per sprite/tile (4-bits) in its 4th generation hardware. But when you use the same speed RAM chips, that's all that you get. You don't suddenly get enough extra RAM bandwidth to display a 2nd (or 3rd) layer of tiles, or put more sprites on the screen.
The only way that you get those extra features is to use faster RAM ... which is more expensive, if it is even available at all.
The PCE was designed to be able to use slow RAM that was the same speed as the NES's ... it's right there in the VDC documentation, register 9 (MWR), which controls memory access width.
As it happens, between the time that the PCE was conceived, and the time that NEC partnered with Hudson to get it manufactured and released, affordable RAM got a lot faster, but we were also going through the fluctuating RAM market that was leading up to Great DRAM shortage. See ...
Out Of Memory Error
tedium.co/2016/11/24/1988-ram-shortage-history/GREAT DRAM GAP WAS THE OIL CRISIS OF COMPUTERS.
www.chicagotribune.com/news/ct-xpm-1988-11-20-8802180788-story.htmlAnyway, at the time that the PCE specs would have been locked-down for release, faster DRAM was available that would have allowed the PCE to do 2-layer backgrounds, but the VDC chip was still light-years ahead of the NES, and so what we got instead was unrestricted access to VRAM during the displayed frame, rather than redesigning the VDC to add a 2nd layer.
As we all know now, both Sega and Ninetendo chose to use the extra bandwidth that was available in the fast VRAM to provide a 2nd background layer and other effects ... but at the cost of locking the CPU out of writing to VRAM during the displayed frame, and so having to rely on DMA during the vblank period to upload new graphics to the VRAM from CPU memory (RAM or ROM).
Having programmed all 3 of the machines ... personally, I'd much rather have the unrestricted access to VRAM than a dedicated 2nd layer of background graphics that is often hidden by the foreground layer.
For me, the knockout graphical effect that seemed to wow game buyers in the 4th generation wasn't the existence of a 2nd background layer, it was Nintendo's Mode7 scaling and rotating effect (which I never personally liked all that much).