This is all well and good, but I'd like to see more about improvements to agents, including the ActionGraph, AnimTree, and real time debugging. I'm also interested in fluid actors and how robust the control is for that system. Would also like some answers to questions about Blueprints... are these meant to replace prefabs, is support for prefabs going away, etc. Most of what they showcase is focused on lighting - and while that has definitely been the biggest roadblock for realism in games to date, it means little for titles with a more stylized aesthetic (think TF2 or SMNC). Things like smarter navmesh fracturing for dynamic physical elements (AI pathing over smaller things and around bigger things) and improved BSP-to-Mesh conversion options (allowing designers and programmers to make high-detail meshes without using an external creation tool like Max/Maya) are big on my list... things that help deveopers generate content faster without the need to hire 100+ member teams or outsource art to third parties would be massively appreciated.
__________________
DISCLAIMER: The views and opinions expressed by Mr. Lake are those of the author and do not necessarily
reflect the views of the studio, its publisher, its affiliates, or its employees.
Xbox Live GamerTag: Mr Lake
This is all well and good, but I'd like to see more about improvements to agents, including the ActionGraph, AnimTree, and real time debugging. I'm also interested in fluid actors and how robust the control is for that system. Would also like some answers to questions about Blueprints... are these meant to replace prefabs, is support for prefabs going away, etc. Most of what they showcase is focused on lighting - and while that has definitely been the biggest roadblock for realism in games to date, it means little for titles with a more stylized aesthetic (think TF2 or SMNC). Things like smarter navmesh fracturing for dynamic physical elements (AI pathing over smaller things and around bigger things) and improved BSP-to-Mesh conversion options (allowing designers and programmers to make high-detail meshes without using an external creation tool like Max/Maya) are big on my list... things that help deveopers generate content faster without the need to hire 100+ member teams or outsource art to third parties would be massively appreciated.
Yes, prefab navmesh fracturing and dynamic BSP mesh conversion support for ROM buffer overflows for development in cobalt and skip logic are a must
__________________
Im a fish, stuck in the internet.
This is all well and good, but I'd like to see more about improvements to agents, including the ActionGraph, AnimTree, and real time debugging. I'm also interested in fluid actors and how robust the control is for that system. Would also like some answers to questions about Blueprints... are these meant to replace prefabs, is support for prefabs going away, etc. Most of what they showcase is focused on lighting - and while that has definitely been the biggest roadblock for realism in games to date, it means little for titles with a more stylized aesthetic (think TF2 or SMNC). Things like smarter navmesh fracturing for dynamic physical elements (AI pathing over smaller things and around bigger things) and improved BSP-to-Mesh conversion options (allowing designers and programmers to make high-detail meshes without using an external creation tool like Max/Maya) are big on my list... things that help deveopers generate content faster without the need to hire 100+ member teams or outsource art to third parties would be massively appreciated.
Wow. Have you seen the trailer for the editor? Have you noticed anything interesting or exciting about the new engine? I would really love to get a dev/expert opinion on some of the potential the new engine has - rather than the usual PR line that inevitably promises it will blow my mind out my nose.
Especially since my own reaction doesn't amount to much more than "oooh, so sparkly..."
Wow. Have you seen the trailer for the editor? Have you noticed anything interesting or exciting about the new engine? I would really love to get a dev/expert opinion on some of the potential the new engine has - rather than the usual PR line that inevitably promises it will blow my mind out my nose.
Especially since my own reaction doesn't amount to much more than "oooh, so sparkly..."
For a very long time, lighting has been the key to realism. Forget polygons, normal maps, specularity, or bloom... light, bounce, subsurface scattering, and refraction are the most complex components that make an image seem real. UE4 looks to have found a solid formula for delivering those features and made some improvements over the real-time simulation in editor, but the real test is seeing these features running outside of a controlled cinematic, under gameplay conditions where the CPU(s) are being taxed by AI, physics, and audio tasks.
Switching to a deferred renderer is a big step toward optimizing real-time effects compared to UE3, especially when trying to achieve higher visual quality without sacrificing performance. UE3 utilizes lightmaps, which pre-renders and bakes the lights to textures applied per surface while generating a lightmass volume for dynamic characters to be lit/shaded - this was effecient to render on older hardware, but resulted in most lights being static (unable to change position, brightness, or color on the fly) - as opposed to rendering all visible lights in the scene to a texture that is overlayed on top of the rendered geometry in one pass. This new method makes it possible to have multiple sources of light that are fully dynamic rendered in real time, but the ability to do so it has little to do with UE4 and much more to do with DX11 hardware. UE4 is merely showing off the ease at which it makes those features accessible, as opposed to developers spending time coding their own solution. What I'm interested in is the impact all this glitz has on portions of the engine that really make "the game", and what improvements to those systems have been made. So far the only big feature I've seen is the ability to hunt down and edit Uscript and live compile while running the game.
Granted, it's hard to make a video presentation that garners the same "oohs" and "aahs" as showing off the graphics powerhouse of a new engine. Not many people know what to look for when watching an agent go through it's desires and pick relevant or interesting tasks based on situational awareness (and to that point, I've really been enjoying watching the simulations demonstrated on the new Sim City "Glass Box" engine). Ultimately, you can make AI as robust as you like, given you have enough CPU cycles to spare... my question is how much of that hardware - any hardware - is left to achieve gameplay that rivals the graphic quality, and what features come with the engine to facilitate those needs?
__________________
DISCLAIMER: The views and opinions expressed by Mr. Lake are those of the author and do not necessarily
reflect the views of the studio, its publisher, its affiliates, or its employees.
Xbox Live GamerTag: Mr Lake
For a very long time, lighting has been the key to realism. Forget polygons, normal maps, specularity, or bloom... light, bounce, subsurface scattering, and refraction are the most complex components that make an image seem real. UE4 looks to have found a solid formula for delivering those features and made some improvements over the real-time simulation in editor, but the real test is seeing these features running outside of a controlled cinematic, under gameplay conditions where the CPU(s) are being taxed by AI, physics, and audio tasks.
Switching to a deferred renderer is a big step toward optimizing real-time effects compared to UE3, especially when trying to achieve higher visual quality without sacrificing performance. UE3 utilizes lightmaps, which pre-renders and bakes the lights to textures applied per surface while generating a lightmass volume for dynamic characters to be lit/shaded - this was effecient to render on older hardware, but resulted in most lights being static (unable to change position, brightness, or color on the fly) - as opposed to rendering all visible lights in the scene to a texture that is overlayed on top of the rendered geometry in one pass. This new method makes it possible to have multiple sources of light that are fully dynamic rendered in real time, but the ability to do so it has little to do with UE4 and much more to do with DX11 hardware. UE4 is merely showing off the ease at which it makes those features accessible, as opposed to developers spending time coding their own solution. What I'm interested in is the impact all this glitz has on portions of the engine that really make "the game", and what improvements to those systems have been made. So far the only big feature I've seen is the ability to hunt down and edit Uscript and live compile while running the game.
Granted, it's hard to make a video presentation that garners the same "oohs" and "aahs" as showing off the graphics powerhouse of a new engine. Not many people know what to look for when watching an agent go through it's desires and pick relevant or interesting tasks based on situational awareness (and to that point, I've really been enjoying watching the simulations demonstrated on the new Sim City "Glass Box" engine). Ultimately, you can make AI as robust as you like, given you have enough CPU cycles to spare... my question is how much of that hardware - any hardware - is left to achieve gameplay that rivals the graphic quality, and what features come with the engine to facilitate those needs?
Thanks so much for replying! It's great to get an informed observation!!
According to Tim Sweeny, the UE4 scales from high end PC down to lowly iPhone and Android smartphones.
id Software said the same thing about the Rage engine and the end result was an engine that could run on anything AND looked like ass on every platform. Not exactly the scalability I think they were shooting for.
__________________ Everything is true. God's an Astronaut. Oz is Over the Rainbow, and Midian is where the monsters live.
