I think yours is probably more accurate. I did my tests in the spawn of a MP server I found, with excessively powerful hardware, and also I never went through the texture packs to check for MCPatcher features.
That said, the packs I used can be taken as representative as the generic 16x, 32x etc pack people would use in practice, because I used the top downloads from Curse. While the fairest 16x pack should have the same performance as Default, by using a pack you're very likely to introduce extra features and animations which must be taken into account when tested from a practical standpoint. Also, using a pack can only keep your FPS the same or decrease it, most likely being the latter, so I'd recommend to anyone who asks to use Default.
To fix the invisible faces in the generated item models, in GIMP do layer > transparency > threshold alpha. It's because you have some pixels that aren't quite erased properly, so the game fails to generate faces where you think the border should be.
I messed around with some monitor settings, but I didn't find a mode that was very good. Maybe it's a mac thing. I didn't know video cards had color blind options
Why would it be better?
Because then it'll apply to all the colors in the game, including text and anything that hasn't been changed/is hardcoded.
It's ok to use the shaders in a complete texture pack without crediting me, but distributing the shaders by themselves is going over the line.
That being said, it is a very grey area, I should have been more clear.
Better send him a PM because I'm sure he has a good reason for whatever he's done.
As shadowforge said, the performance drop is mostly determined by how many MCPatcher features and animated textures the pack includes. (Also, stereo, long-length, or poorly optimized custom sounds could cause some drops as well).
Interesting! Were you using MCPatcher or Optifine?
Odd that there's always a drop from the vanilla framerate... I guess that would confirm what SirTulip had heard.
This was with Optifine because I think that anyone who this graph interests should be using Optifine anyway. The variance during my tests was larger than I'd like especially getting up to the larger resolutions where not all the packs were complete or used special features, but I think it's pretty clear that default is the best. I will have to do more tests later on a lower-end computer.
That's pretty much all you can do. It will look grainy, so it's best to play in a high resolution so each pixel is less significant. Also you know how it'll have a solid inside and a grainy halo (maybe yours won't, but mine definitely does)? If you maximize the inside and make the halo smaller, well the inside is smooth so it'll look smoother close to the blade.
I have two planes (0-width cubes, just checking) on either side of the blade, about 0.001 pixels out to prevent z-fighting. Both planes have two sides textured.
0
I think yours is probably more accurate. I did my tests in the spawn of a MP server I found, with excessively powerful hardware, and also I never went through the texture packs to check for MCPatcher features.
That said, the packs I used can be taken as representative as the generic 16x, 32x etc pack people would use in practice, because I used the top downloads from Curse. While the fairest 16x pack should have the same performance as Default, by using a pack you're very likely to introduce extra features and animations which must be taken into account when tested from a practical standpoint. Also, using a pack can only keep your FPS the same or decrease it, most likely being the latter, so I'd recommend to anyone who asks to use Default.
I think more testing will be required
0
1
Because then it'll apply to all the colors in the game, including text and anything that hasn't been changed/is hardcoded.
1
0
Better send him a PM because I'm sure he has a good reason for whatever he's done.
1
This was with Optifine because I think that anyone who this graph interests should be using Optifine anyway. The variance during my tests was larger than I'd like especially getting up to the larger resolutions where not all the packs were complete or used special features, but I think it's pretty clear that default is the best. I will have to do more tests later on a lower-end computer.
0
1
Quick test I put together with 3 packs at each resolution (except for default and 7x) averaged over 5 samples a minute apart.
0
0
0
0
0
0
0
http://imgur.com/a/A8Vg1