I do think there's still a big gap between stuff like maxed out Alan Wake 2 and full on photorealism but peak real time graphics are getting close enough to it where diminishing returns are getting real. Future engines like UE6 will help with the budget issue but starting next decade getting closer to simulating real life in other aspects will become more important than simulating it visually. A game with not quite photorealistic but still good enough visuals and an extremely believable world chock-full of interactive elements and compelling NPCs will be more immersive for probably most people than one with photorealistic visuals but a dull world with not much interactivity.

Resolution itself doesn't actually increase budget costs.
I could grab a PC game from 1998 and run it at 16k resolution with the press of a simple button, it's development budget would probably be comparable to an indie game today.

It's making all the high quality assets that costs money... And it doesn't matter if it's 720P, 1080P, 4k or 8k, we aren't photo-realistic at any resolution yet.

We need better developer tools and technologies to streamline game development... And things are improving there.
I.E. Ray Tracing removes the burden from artists to make baked lighting/shadow and let the graphics engine handle it in real time and dynamically, that actually can save money outside of cutscenes which require specific light and shadow placement to obtain a certain look.

Lets just by magic you had a console-viable chipset that could render even Avatar 2 type visuals in real time for $600. You're not magically going to get movie quality graphics like say Avatar 2 in real time without there being a massive cost behind that (1 billion dollar budget? 2 billion? Unlike movies games have to have full environments, you can't just cheat whatever is in the frame).

In the relative future that just isn't realistic, even if the hardware existed to do that in real time (it doesn't, not even close), developers don't have 1-2 billion + 10+ year dev cycles it would take to even use such hardware properly.

And honestly most people don't even care, which makes the whole pursuit even more pointless.

Yet PSVR1 to PSVR2 is a massive generational leap. And VR has room for many more leaps in resolution, fov, comfort, full body tracking, variable focus (solving vergence-accommodation conflict), games.

Console games can also still advance. Living worlds is a dream that's still in its infancy. A-Life 2.0 in Stalker is where the advances are at. (Remains to be seen if it actually works, not yet) Physics still have miles to go before they no longer have to be scripted.

It's just graphics that don't visibly advance that much anymore, or don't make that much of a difference anymore. Hence future consoles should focus more on CPU cores and RAM to make other advances possible. Physics based worlds instead of model sets.

Yet the industry still sells games based on screenshots and videos, eye candy. Consoles are powerful enough to explore more games like From Dust, yet people still flock to AAA eye candy instead of lesser looking games with more interactivity.

As long as people keep pre-ordering AAA eye candy en masse, why would the industry focus on the gameplay :/