Men's Round 5 Finish - Invisible Rider? Confusing Finish

(Mike ) #1

Some odd happenings going on with the finish last night. Still issues with camera switching which made it confusing as ■■■■. I’ve watched it 5 times to get my head round things. Which also led me to noticing the following oddness.

As you can see in the screenshot below Ian Bibby (I think, please sort out team names) is in second behind C.Brown.

Yet just a couple of seconds later he has disappeared. Possibly swamped by other riders but the time gaps don’t make sense

Cut forward 5 seconds and A.West is now clearly in front of Brown. The names haven’t switched places yet (lag?) on the right but Brown is now shown as being in 3rd place. Has he been overtaken by an invisible Bibby? Is this the most powerful Power-Up there is? Stealth Mode!

Then we cut to the finish line and suddenly Bibby is visible, just behind West taking the win. As we are following the lead rider we then miss Jon Mould taking Brown on the line for 3rd. Zwift please sort out some static cameras for the sprints and finish line.

I’ve re-watched this a few times, not becuase I’m trying to finding faults with Zwift to bash them with but because I was really confused when I watched it live and what you can see, or rather don’t see is Bibby overtake Brown. In the final few hundred metres Bibby drops off the rider list, disappears and then reappears just before the finish line. At some point I think West must have overtaken Bibby too. Let me make it clear at this point that I am in no way accusing Bibby of manipulating the software or cheating and I am not looking to tarnish his reputation in anyway at all. It’s unfortunate that it happens to be him but could be anyone.

So what I am saying? Well firstly, how can I believe the result? With riders popping into and out of view it makes it impossible to follow the finale of the race. I’m new to Zwift racing so I don’t really have any affiliations yet but imagine you are with Brown, cheering him on do you think he has got 2nd? 3rd? 4th? You just don’t know. DC Rainmaker has written a very good post about some of the limitations of Zwift racing looking at hardware and rider validation but he does not really touch on the validity of the software platform itself. I understand this first season is a proof of concept season but I have also read that there is prize money at stake, as well as pride and prestige of course. You better be damned sure of the results if handing out large sums of cash.

Then there is the fan experience which I slightly touched on. The end of the race was confusing, underwhelming and in places missing. Why have sprint and KOM competitions if you’re not going to show the results at the end of the race? Also, why, 12 hours after the event are there no results on the Zwift website, let alone a write up. It will probably follow some time today but this isn’t the 1930’s when you had to wait until the next days newspaper to get the results. At least Zwift are putting something on their website - this is the SuperLeague page on the KISS website.

This post is long and in parts negative but Zwift are putting money into this, which could be spent on game development. It’s a cycle though, raise the profile of the platform to attract more investment which benefits everyone, so it’s important that Zwift get feedback in order to optimise this new era of eSports and to make sure that it doesn’t fall flat on its face and step number 1 is to give fans confidence in the results.

(Daren) #2

Camerawork

Although many aspects of the race coverage have got better over the past month, camerawork has consistently let us down.

Sometimes even failing to get a camera on the very start of the race, so we don’t see the initial huge surge as the barrier disappears. It happened last week and again last night.

Annoying, but OK, only the start of the race.

However, several times last night we missed or got a poor view of key moments. An example of last night was the first time of the Richmond KOM.

The camera was following Ollie Jones, so we didn’t see Lionel Vujasin actually take the KOM. OK, you could argue that he was far ahead and the 2nd/3rd place battle was more interesting. But even that suffered. Here you can see how close it was:

Close, right! But here are two problems:

  • there’s no actual line on the road to mark the winning point. I’ve added one here as an approximation of how close it was.
  • the camera cut to Vujasin immediately after this screenshot. So we didn’t actually see them cross the non-existent line anyway.

Watch it in slow motion to see what I mean. (https://youtu.be/uhKvFoqeEXk?t=1498) By the way, Nathan sounds a lot less hyper at 0.25 speed. :smiley:

At other times we end up following a rider as he gets swarmed (or even just passed by a rider or two). But rather than switching to the new leader, the camera lingers on the rider who’s been passed. So we miss a lot of the real drama at the cutting edge of the race.

I’m sure the Zwift team know camerawork needs improvement. Whoever was hosting the YouTube channel last night said as much. But it absolutely must get better, because being able to follow the action is fundamental to enjoyment of sporting coverage.

Disappearing riders

Riders disappeared from view quite a lot last night. It’s jarring and undermines confidence.

Anyone who has spent time on Zwift has probably experienced this first-hand. Occasionally people disappear from the game, then reappear a bit later. Sometimes everyone disappears, then comes back. This is normally attributed to dropouts.

OK, there may be occasional dropouts between ANT+ or BT equipment, but modern internet connections themselves are generally very stable things. When I experience everyone disappearing, I don’t think it’s my connection. First, my power and cadence data is still being sent to Zwift. Second, other internet services are continuing to function (e.g. Spotify streaming).

So I do wonder if there are server-side reasons for people dropping out.

Whatever the reason, riders simply mustn’t just disappear from the screen in race coverage. At worst the client should extrapolate or predict based on the last known position and power of a rider. There might need to be some readjustment when they come back, but is that worse than them just vanishing then popping back in from nowhere?

Multiplayer games, particularly first-person shooters have had to deal with these sorts of issues for decades, so there’s a lot of expertise out there about how to improve network quality and the player experience.

Server wins, client view is absolute

In multiplayer games one of the questions that must be addressed is who has the authoritative view of the game world?

If it’s a client wins scenario, then if I shoot someone in the head on my screen, then that it what happened. That then has to be reconciled with what another player sees, and can lead to problems when they feel they were around a corner due to latency etc.

If it’s server wins, the even if I think I’ve made that headshot, the server can overrule my client and say “Nope, you missed.”

Anyone interested can find many articles and videos on this topic. e.g. https://gafferongames.com/post/what_every_programmer_needs_to_know_about_game_networking/

So for various reasons the server must be the final authority as to who crossed the line first.

But we must absolutely be able to trust what we see on our screens. If we see that Player_A crossed the line half a tyre’s depth ahead of Player_B, that must be what happened. It’s unacceptable to later be told “ah, no, official results say Player_B won.”

In the real world the speed of light is pretty quick, so what we see on TV can be relied upon. We never see one person cross the line first only to be told they were actually second. (Photo finish cameras notwithstanding, but they only tend to come into play whern things are too close to call).

The client should only display what it knows to be true - and that means all riders all the time, as well as their accurate positions.

There may be technical limitations, but in that case I don’t see how this sort of racing can really take off as a spectator sport. I dare say there are various online racing games that have tackled this issue e.g. Assetto Corsa in the car sim racing world.

As Mike said, this does sound pretty negative. It’s intended as constructive criticism, borne from a desire for Zwift racing to really take off and succeed with mass appeal, reliability, great entertainment and reliable “what you see is what happened” coverage.

Development takes time, and new features need prioritisation. I understand that. Not expecting magic wands or anything, but a commitment, publicly stated, to address these issues I guess.

2 Likes
(Mike ) #3

Part of the issue seems to be the tools provided to the ‘broadcaster’ - which I think is being outsourced by Zwift. They seem to be the same as any general Zwifter would have, so you can only pick camera position by following a particular rider. I suspect this is why it’s so difficult to for the person choosing the camera angles to follow the sprints.

(Daren) #4

Yeah, I’m sure it’s essentially the default game client.

Maybe a branch with a few tweaks in (such as the handul of additional overlays we’ve seen - distance to go, “Breakaway” etc.). But ultimately no better than one of us logging in to the game and spectating riders ourselves.

In fact it’s worse, since we have no control over who is followed or the camera angle. The race coverage could be improved easily by enabling us to join in an enforced “just watch” watch mode and choose our own coverage. We can still listen to Nathan’s commentary.

2 Likes
(Mike ) #5

I only use Zwift through an iOS device so as far as I know I’m not able to watch riders but I understand that you can watch any rider you follow on Zwift if you are using the PC/Mac version of the software, so it does beg the question, why not just do that and control the view myself. If I’m not getting a curated, smooth experience with additional info that might well be preferable, with the live stream in the background for the commentary.

(Vincent) #6

I wonder about internet connection issues in some of these cases.

Pink kit in the sprint finish disappears at 43:36 and then reappears right before the finish line to take the win?!?!?!? When he disappeared from the screen his name also disappeared from the leader board on the right. Name reappeared when he did just before the finish.

I am guessing there was some lag or connection drop, etc?

(Johnathan) #7

Don’t worry, MGF takes appropriate measures are taken to insure the most correct results possible are published.

However disappearing riders is a known issues with Zwift. Watch any KSL live coverage, riders pop in and out most of the race. Thankfully I think the riders that were disconnecting their internet to seek an advantage have stopped that since they were caught a few months ago.

(Vincent) #8

I am not really sure but don’t you lose any benefit of a draft when you disconnect? That would mean that the only place where it may be an advantage is the final sprint. There is no way you are going the long haul without a draft. Lastly if you disconnect you disappear for everyone else but doesn’t everyone else disappear on your screen? How would you even know where you stand? Again maybe not a huge deal in the final sprint but that seems like critical info to not have in the middle of a race.

Bottom line is that I don’t see how intentionally disconnecting is of any advantage but I have never done it and maybe I am missing something here…

(Johnathan) #9

A few months ago, a couple riders figured out that they could disconnect, attack and get an advantage without the group knowing, then reconnect and win the race.

To my knowledge, this is why you can see rider’s lag on ZwiftPower.

1 Like
(Mike ) #10

Interesting I didn’t realise you could see lag. Looking at those figures the lag is greater than the margins for some of the race wins. I wonder if that is affecting results.