oops read that wrong #3 is correct

cl_cmdrate 66

How new change is this? Since my friends can get their interp to 0.0152

If you have a weird cl_imterp_ratio that could mess with it (use 1)

as #3 said, make sure your cl_cmdrate is 66, if is isnt your cl_updaterate probably isnt either. Set them both to 66

My ratio is set to one also

Is this supposed to replicate the lower interp or should it make it possible to change interp it 0.0152?

The higher cmd, the lower your interp can go. TF2 maxes at 66 so the lowest can only ever be .0152. However, your cmd rate apparently is 20 since you can't go lower than .05. For some reason, this happens to me. Once of the reasons I keep a constant check on my interp.

If you make cmdrate 66 you will be able to drop to your desired interp. However, also make cl_updaterate is 66. That will help your game in general.

Now it's working, thanks a bunch!

It's actually cl_updaterate that affects interp, but cmdrate doesn't have a reason to be lower than updaterate, so it's set to the same thing.

you can actually go below 0.0152 interp

http://i.imgur.com/TCoh4qd.png

^ screenshot taken in firepowered MGE for convenience sake, but I haven't seen my interp go above 0.1 in months
im currently using these rates

cl_cmdrate 100
cl_interp 0.01
cl_interp_ratio 1
cl_lagcompensation 1
cl_pred_optimize 2
cl_smooth 0
cl_smoothtime 0.01
cl_updaterate 100
rate 80000

and despite servers being limited to 67 cmd/update rate, the server still shows that im at 0.01 interp despite console saying my cmd and updaterate have both been set to 67 by the server

I dont really understand why it happens, but this seems to be the lowest possible in tf2, as anything above 100 just doesnt work.

doesn't that just goof up your reg and make models really jumpy? that's what should happen if your interp is too low for your rates lol

no, its perfectly fine for me, and i havent noticed tf2s reg being any worse than usual

Downpour, the lowest achievable interp in tf2 is actually 0.0152 ms, because your cmdrate and updaterate won't go higher than 66, since all servers run at a max of 66 ticks.
The reason why it shows you 0.010 lerp is that the net graph is bugged and you can fool it into thinking that both of your rates are at 100 even though your game only runs on 66 max.
For example, set both rates to a million and interp/ratio to 0, it will show you 0 lerp even though that's impossible.

Also set your rate from 80000 to 60000, your game won't use 80000.

here's something someone smarter than myself made and was on a not-so convenient site:
http://www.konr.co.uk/lerp

Konr, that's a link to an article on Left 4 Dead networking - A 30 tick game.

It conflates ping and interpolation in the first section, so I'm not sure it's valuable to anyone.

If you want solid information on Source networking, look here:

https://developer.valvesoftware.com/wiki/Source_Multiplayer_Networking

Anything I've read elsewhere takes the information found on those page and mutates it into advice that often does more harm than good.

Isn't this a projectile interp though? having some interp as hitreg can be good.

i dont have enough knowledge of the netcode of tf2 to argue this, but i dont know if ill take it as fact.

doesnt really make a difference imo

trust me, all the tricks to get below 0.0152 lerp are pure placebo

There is no benefit to reducing interp below 0.0152. Even if it works, you're just going to get extrapolation (jittery models) and/or bad reg. There's no real benefit to the extra 5 milliseconds when this game has lag compensation.

Here's something I wrote on interp way back when. It's pretty dense, so maybe someone will rewrite it or quote out only the most important parts. Go ahead if you want to do so, just don't quote anything so out of context that it's wrong. https://dl.dropboxusercontent.com/u/1811521/interp.html

tf2_scientist: lol i knew that would happen
tf2_scientist: you just give them the rates i gave you with no explanation
tf2_scientist: and you can't explain it so they don't believe you
tf2_scientist: there's a more important part to it beyond the -5ms anyway
tf2_scientist: also that german guy is retarded
tf2_scientist: if you set your updaterate to like 1000 or 10000
tf2_scientist: you actually will get 0 lerp
tf2_scientist: but the game will be really glitchy and models really will warp
tf2_scientist: at 10ms they don't warp for me at all
tf2_scientist: it's like in l4d
tf2_scientist: servers were 30 tick but you could force your interp lower than 67.0 by using a higher updaterate
tf2_scientist: and it definitely worked because when you set it lower, common infected would be jittery
tf2_scientist: but it was worth
tf2_scientist: this is what happens when you have a bunch of mouthbreathers passing on the same shit for years

If you just substitute the L4D info he's referencing with TF2 info you get better information than most places. The game works the same way (at least in the examples he gives) with the exception of the tickrate. Not a difficult distinction to make!

So because the game is glitching out with the information it's trying to use to draw models etc that means it "worked"? Lmao..

tf2_scientist: models being jittery is a byproduct of having a lower interp like everyone else said
tf2_scientist: so that's exactly what it means
undeniably, love: it means that the lower interp must be happening
tf2_scientist: ^
tf2_scientist: and the point is
tf2_scientist: they would be visibly more jittery when using a higher updaterate to force a lower interp (an updaterate above 30)
tf2_scientist: updaterate, cmdrate, and interp are clientside
tf2_scientist: you can make your client do whatever you want regardless of what the server sends you
tf2_scientist: except there are clientside caps on some things in some games
tf2_scientist: for example, the 100 cmdrate cap in tf2 and l4d

It's a byproduct of the settings you have but it doesn't prove that the interp is actually working at the settings that are shown on net graph at all.
EDIT:
If it was truly working as you think it is then the jittering wouldn't happen because the settings that the game is actually using and the settings that net_graph is saying are being used wouldn't be different. That among other things is why I imagine the game jitters on these settings because the actual lerp is not at all what the game is suggesting.

The engine is going into extrapolation mode, which means that things are going to be ahead of where they actually are. You might think this is a good thing, but in reality, it doesn't mean you can just aim behind things and hit them unless they're moving in perfectly straight lines at a set speed. In reality, you're going to get people moving on curves and stutterstepping, and extrapolation basically puts the model in entirely wrong places over time in those cases.

Have a visual aid:

http://i.imgur.com/BPIeVzl.png

If a person already knows the network specifics of TF2 (which they would need to swap out the information, as you suggest), then why would they be looking at such an article, let alone one written specifically for Left 4 Dead networking?

If someone unfamiliar with the networking of TF2 were to read that article they would be either be seriously confused by all the very Left 4 Dead specific details in that article, or would end up with some very unhelpful network settings.

There's all kinds of misleading information and advice in that article, but my favourite was one he chose to write in bold:

If you have yellow lerp, asking the server admin to turn up the server framerate is a good idea.

Good luck asking your server admin to turn up the framerate in your next game!

For what it's worth, wareya's article is pretty solid.

jesus christ just play with whatever update rate, cmd rate and cl_interp you want and stfu.

The OP's question has already been answered and you tulips are still in here bitching about a subject that's been beaten 98389248972389243 times more then the fucking dead horse... go find something else to do and chill.

sorry, i was asleep

LOL, opinion discarded.