BladeW
Member
After Epic uploaded the Unreal Engine 5 demo video, the heated discussion of Unreal Engine 5 hasn been flooding the screen. If you watched the demo, you will marvel at the Nanite virtual micro polygonal geometry which allows artists to create all the geometry details that the human eye can see.
The emergence of Nanite virtual geometry means that film and television level art works composed of hundreds of millions of polygons can be directly imported into the virtual engine - whether it's sculptures from ZBrush or CAD data scanned by photogrammetry. Nanite geometry can be streamed and scaled in real time, so there is no need to consider the budget of polygon quantity, polygon memory or painting times; there is no need to bake details to normal map or edit LOD manually, and the picture quality will not be lost. Then I watched the demonstration video... It's worthy of being the next generation of game screen, surpassing the existing game screen. But from the perspective of the industry, the quality of the screen is still secondary, and the most important is the production process
In the current industry process, after a model is made, the art needs to make a low-precision version, normal map and baking static light map... But this is not the longest step of the process. The longest step of the process is usually to find that the number of faces of the model is too much in the actual measurement of the game, and the draw call and memory occupation are too high, so the previous process needs to be repeated, Reduce the number of faces, adjust the LOD, skillfully mix the static light and real-time light and shadow, and finally achieve the balance between performance and picture quality.
Art resources are very expensive, and a large part of money is used to "adjust (reduce) accuracy" and "auxiliary production resources". Now Unreal Engine 5 says that this set of flow does not need... After the model is finished (or scanned), just put it in directly... 100 million faces or one billion faces, You don't have to care... That's why we call out for miracles.
From the perspective of production line, the popularization of this technology will greatly reduce the production cost and process of art resources, and liberate a lot of manpower from the tedious work. Exaggeration may reduce the demand market of some manpower. How does the miracle come true? There is not much information disclosed now, so we can only make some guesses. Most people do not believe it, In the case of no miracle in hardware, rendering performance can be improved hundreds of times by engine algorithm alone, so many people speculate that the virtual engine automatically processes these resources, automatically generates resources of appropriate accuracy in advance according to the performance of the target machine, and switches in real time in the game. This guess is also confirmed in the video.
But wait... 20 million? This is still an order of magnitude higher than the number we mentioned in our daily life... How to improve this order of magnitude? You seem to have no idea for the time being... Correction: some people point out that it can realize tens of millions of faces on the PC now, so this data can understand that PS55 boasts that all technologies are provided by the virtual engine, However, it is repeatedly emphasized in the video that the real machine image is demonstrated on PS .
Although it is certain that all the mainstream platforms in the future, including Xbox, PC and mobile phone, can enter the next era in succession, at present, PS5 is probably the most appropriate demonstration platform. After all, although PS5 is not good at anything else, the SSD architecture is very amazing. Such an excellent picture, With such smooth dynamic performance, everything can be optimized. Only the size of data can't be optimized: there's no ultra-high speed hard disk reading speed, I'm afraid I can't enjoy this feast of the times.
At the beginning of the year, the first show of PS5 was criticized by many players: almost half of the time, I talked about the hardware architecture principle of that SSD, without any real pictures, The published GPU parameters are also unsatisfactory. But looking back, maybe Sony's layout is quite profound: the next generation of game engine will automatically handle the global dynamic light and pixel level number of faces. The biggest bottleneck is in hardware transmission. If Sony had published this real video of PS5 earlier that time, Then PS followers won't have to bear the humiliation for so many days.
Looking forward to the future, although there are real machine demonstration videos and full of big talk, Unreal Engine 5 is still a long way from us. The preview version of Unreal Engine 5 will be released in early 2021, and the full version will be later. Such excellent performance, even for all hardware vendors, including computers and mobile phones, also requires corresponding time to adjust and adapt. In the next year (or several years), It's time to see what's next.
The emergence of Nanite virtual geometry means that film and television level art works composed of hundreds of millions of polygons can be directly imported into the virtual engine - whether it's sculptures from ZBrush or CAD data scanned by photogrammetry. Nanite geometry can be streamed and scaled in real time, so there is no need to consider the budget of polygon quantity, polygon memory or painting times; there is no need to bake details to normal map or edit LOD manually, and the picture quality will not be lost. Then I watched the demonstration video... It's worthy of being the next generation of game screen, surpassing the existing game screen. But from the perspective of the industry, the quality of the screen is still secondary, and the most important is the production process
In the current industry process, after a model is made, the art needs to make a low-precision version, normal map and baking static light map... But this is not the longest step of the process. The longest step of the process is usually to find that the number of faces of the model is too much in the actual measurement of the game, and the draw call and memory occupation are too high, so the previous process needs to be repeated, Reduce the number of faces, adjust the LOD, skillfully mix the static light and real-time light and shadow, and finally achieve the balance between performance and picture quality.
Art resources are very expensive, and a large part of money is used to "adjust (reduce) accuracy" and "auxiliary production resources". Now Unreal Engine 5 says that this set of flow does not need... After the model is finished (or scanned), just put it in directly... 100 million faces or one billion faces, You don't have to care... That's why we call out for miracles.
From the perspective of production line, the popularization of this technology will greatly reduce the production cost and process of art resources, and liberate a lot of manpower from the tedious work. Exaggeration may reduce the demand market of some manpower. How does the miracle come true? There is not much information disclosed now, so we can only make some guesses. Most people do not believe it, In the case of no miracle in hardware, rendering performance can be improved hundreds of times by engine algorithm alone, so many people speculate that the virtual engine automatically processes these resources, automatically generates resources of appropriate accuracy in advance according to the performance of the target machine, and switches in real time in the game. This guess is also confirmed in the video.
But wait... 20 million? This is still an order of magnitude higher than the number we mentioned in our daily life... How to improve this order of magnitude? You seem to have no idea for the time being... Correction: some people point out that it can realize tens of millions of faces on the PC now, so this data can understand that PS55 boasts that all technologies are provided by the virtual engine, However, it is repeatedly emphasized in the video that the real machine image is demonstrated on PS .
Although it is certain that all the mainstream platforms in the future, including Xbox, PC and mobile phone, can enter the next era in succession, at present, PS5 is probably the most appropriate demonstration platform. After all, although PS5 is not good at anything else, the SSD architecture is very amazing. Such an excellent picture, With such smooth dynamic performance, everything can be optimized. Only the size of data can't be optimized: there's no ultra-high speed hard disk reading speed, I'm afraid I can't enjoy this feast of the times.
At the beginning of the year, the first show of PS5 was criticized by many players: almost half of the time, I talked about the hardware architecture principle of that SSD, without any real pictures, The published GPU parameters are also unsatisfactory. But looking back, maybe Sony's layout is quite profound: the next generation of game engine will automatically handle the global dynamic light and pixel level number of faces. The biggest bottleneck is in hardware transmission. If Sony had published this real video of PS5 earlier that time, Then PS followers won't have to bear the humiliation for so many days.
Looking forward to the future, although there are real machine demonstration videos and full of big talk, Unreal Engine 5 is still a long way from us. The preview version of Unreal Engine 5 will be released in early 2021, and the full version will be later. Such excellent performance, even for all hardware vendors, including computers and mobile phones, also requires corresponding time to adjust and adapt. In the next year (or several years), It's time to see what's next.
