Back to the road · README

Road Rash: collisions

What happens when a bike touches something, gathered after riding through other bikes showed the earlier research had never asked. Frame-sheet and text-dump names refer to local captures of the sources listed in the ledger; the captures themselves are EA's and are not redistributed.

Companion to road-rash-research.md (mechanics) and road-rash-visuals.md (look). This file covers only what those two leave out: what happens, frame to frame, when the player's bike touches something, and how the game shows it.

Evidence base: the three gameplay videos in rr-img/_video/ (pc-all-levels.mp4 = 1280x720 "UnbiasedPlayer" HUD, cited here as PC1, same source as YT1 in visuals.md; pc-win95-papyrus.mp4 = 640x360 "Player1" HUD, cited as PC2 / YT2; 3do-longplay.mp4 = 640x480, cited as 3DO), reviewed as contact sheets in rr-collide/ (file + grid position given per fact, e.g. pc1_D_crash_napa.jpg r3c3), plus pc_manual.txt, 3do_text.txt/3do_djvu.txt, the three GameFAQs text dumps (one for the Genesis original, one for Genesis Road Rash 3, cited as RR3 FAQ), and rv_all.txt (cached GameFAQs/review pages, cited rv). A Chinese web search for 撞车/撞人/追尾/被撞飞 (4 Oct 2026) turned up only the Chinese Wikipedia plot/mechanics summary and gamersky's listing page — nothing with crash-specific detail beyond what the English sources already give; it is not cited further below.

A provenance note on the frame set: rr-collide/scan/d3_s*.jpg and d3_K_ped.jpg were unexplained in the handover (no matching resolution/HUD among the three known videos). Viewing d3_s01.jpg resolved it: the sheet opens on a 3DO console being powered on and a disc inserted, then the EA/3DO boot splash and the Road Rash menu (Restroom → Axle select → Pacific Highway/The City/Napa Valley/Der Panzer Klub) — i.e. this is the start of 3do-longplay.mp4, extracted at a different crop/scale than the f3do/ frames used elsewhere, not a different or later game. It is cited below as 3DO like the rest of that file.

1. Rider vs rider — side contact ("rubbing")

• Riding alongside an opponent and leaning into them pushes the other bike sideways without either rider going down; this is also what the kick move does on purpose ("the kick shoves the target's bike sideways", already in road-rash-research.md §3.4). The RR3 FAQ confirms the shove-only outcome is the general rule for lateral contact, not just the kick: "Will push a biker aside. Not very useful. You can get the same effect by bumping into another biker." (RR3 FAQ, _text/gf_genesis_586427-road-rash-3_faqs_29823.txt L150-151)
• At a steeper angle or higher closing speed, a side rub does put the other rider down rather than just shoving them: in PC1, the player closes on a pink/purple rival on a bend in the city; the rival goes into a sideways lean/slide (still half-mounted, not a full ragdoll throw) immediately next to the player, while the player rides on unaffected and the HUD lap digit ticks over. (observed in video — pc1_A_detail.jpg rows 4-5; corroborated by the wider sequence in pc1_A_crash_city.jpg and pc1_A_after.jpg, where the downed rival is passed by several more racers over the following seconds before the sheet ends, i.e. getting up is not shown within ~6 s)
• So there appear to be two distinct rider-down animations depending on how the contact happened: a lean/slide for a side-angle knockdown next to another rider, versus the full spread-eagle "thrown clear" ragdoll seen after hitting a car or fixed obstacle head-on (§3 below). (observed in video, inferred from comparing pc1_A_detail.jpg to pc1_D_detail.jpg)
• Player speed in the PC1 side-rub sequence does not visibly drop on the HUD dial across the contact (needle position unchanged frame to frame); the cost of this kind of contact lands on the other rider, not the player. (observed in video — pc1_A_hud.jpg)
• Being knocked off yourself works differently from crashing yourself. The RR3 FAQ is explicit that inertia differs: "When an enemy biker hits you off your bike, your bike keeps moving forward but you fall off. Unlike a normal crash — where both you and the bike have the same inertia — the trip back to your bike is twice as long... It's actually safer to crash into stuff sometimes [than to be knocked off by another biker]." (RR3 FAQ, L347-352)
• The same FAQ also states knockdown-by-opponent costs no damage-gauge hit by itself: "When you are knocked down by an opponent you take no damage." (RR3 FAQ, L101) — this sits alongside, not against, road-rash-research.md's "yours drains as you take hits" (stamina, not the bike-damage gauge, is what punches/ kicks drain; losing the bike to a knockdown is a time cost, not a gauge cost). Marked (FAQ, RR3) since this text dump is for the Genesis Road Rash 3, not the PC original — treat it as evidence the knockdown/crash split exists in the series, not as a confirmed PC-exact number.

2. Player rear-ends a rider / is rear-ended

• No frame set captures a clean rear-end between two bikes specifically (as opposed to a bike overtaking and rubbing alongside, §1). (clearly unanswerable from the current frame set — would need a fresh extraction targeting a moment where the player is directly behind a slower rival on a straight, which was not found in the batches reviewed)
• The manual's closest analogue is for rear-ending slow car traffic, not another rider, and it explicitly treats rear strikes as just as damaging as head-on: "rear-ending grandpa can be almost as damaging as a head-on collision" (Napa Valley track blurb, pc_manual.txt L515-516, manual). Lacking a rider-specific source, this is carried over as the best available evidence that the game does not treat a rear hit as "free" the way a glancing side rub can be.

3. Hitting a car — same direction

• Confirmed, high-confidence sequence: PC1's Napa Valley run shows the player closing on a blue pickup truck and white sedan sharing the player's lane, then impact — two frames show a pixelation/glitch-style effect consistent with the engine's impact flash — immediately followed by the rider shown airborne in a spread-eagle "thrown clear" ragdoll pose above the road surface. (observed in video — pc1_D_crash_napa.jpg, pc1_D_detail.jpg)
• Speedometer: the paired HUD-only sheet for this run shows the lap/position digit cycling 13→12→11→12→13 through the sequence and the small directional counter moving 00.3→00.4, i.e. a real but not dramatic change in that counter; the main MPH needle itself is not legible at contact-sheet resolution in this set, so an exact before/after MPH cannot be read off (observed in video — pc1_D_hud.jpg — but marked unresolved: needs a fresh higher-res extraction of just the needle if an exact number is required).
• After the ragdoll lands, the rider lies on the road while the field passes, then gets up and runs (pc1_D_after.jpg: rider upright and jogging, HUD rival name cycling Satish→Sven→Ikira→Jorg→Axle over the following seconds) — this matches road-rash-research.md §4.2's "the rider runs back to the bike" almost exactly, now confirmed on video rather than only from text. (observed in video)
• This is the clearest evidence for "hitting a car throws the rider off and separates them from the bike", already stated in road-rash-research.md from text sources; video adds the ragdoll pose and the lie-down-then-stand-and-run beat, plus that nearby racers do not stop or swerve for the downed rider — they ride straight past. (observed in video)

4. Hitting a car — oncoming

• Not captured cleanly in the reviewed batches: every car seen in a crash sequence (pc1_D, pc1_F, pc1_G) is travelling the same direction as the player in the player's lane, not oncoming in the opposing lane. (clearly unanswerable from the current frame set without a new targeted extraction)
• Indirect manual evidence that oncoming impact is modelled as more severe than a same-direction rear hit, by the explicit comparison already quoted above: "safer riding in the left-hand lane than in the right... rear-ending grandpa can be almost as damaging as a head-on collision" (manual, pc_manual.txt L514-516) — implies head-on is the worse end of the scale, with rear-ending slow traffic deliberately called out as surprisingly close to it rather than obviously milder.
• The kick move is explicitly designed to knock a rival into the opposing lane: "one well-timed kick can knock 'em into the oncoming traffic" (already quoted in road-rash-research.md §3.4, manual) — i.e. being knocked into oncoming traffic is a named, intended hazard, but neither the manual nor the reviewed frames show what then happens to the oncoming car or the knocked rider at the moment of that second impact. (unverified)

5. Pedestrians

• 3DO: the track-select blurb for The City (seen on the menu screens at the start of 3do-longplay.mp4) names pedestrians as a track hazard directly: "Dodge nasty traffic and zombie pedestrians in the urban jungle." (observed in video — scan/d3_s01.jpg, menu screen; the word "zombie" here reads as the game's own slang/flavour text for jaywalking pedestrians, not a literal enemy type — no undead pedestrian is seen in any reviewed gameplay frame)
• 3DO gameplay: later in the same run (scan/d3_s15.jpg), at mph in the 130s with HUD rival "Cydney", a pedestrian figure in white appears directly in the player's path with a raised-arm pose consistent with flinching or being struck, immediately followed a few frames later by another rider (rival "Slash") going down nearby — but the sheet's low frame rate does not let me confirm whether the pedestrian or the rival crash caused the other, so this is marked (observed in video, ambiguous) rather than a clean causal fact.
• PC1 coastal/cliff sequence: a pedestrian stands beside the nearly-stopped player bike (speed reading near 15/00.0) and changes pose across frames in a way consistent with a punch/kick exchange rather than a vehicle strike — i.e. this reads as the player fighting a pedestrian on foot after already being stopped, not a moving-bike-vs-pedestrian impact. (observed in video — pc1_C_ped.jpg)
• Manual: pedestrians are listed as a named hazard of The City course alongside traffic and buildings ("Commuters, pedestrians, and buildings... make this one of the most challenging courses", pc_manual.txt L487-489, manual) but the manual does not give a damage number for hitting one.
• RR3 FAQ gives a hazard-tier breakdown that is the single most concrete thing found on contact severity, albeit for animals rather than pedestrians and for a different game in the series: "stuff such as the chickens in Italy or the kangaroos in Australia can be ran over without damage, but stuff like cows and zebras will cause damage and crash you." (RR3 FAQ L99-101) This supports road-rash-research.md §2.5's point that obstacles are sorted into harmless/ jump/crash tiers, now with the concrete rule that tier depends on the specific object, not just size — extending it to say the same is plausible for pedestrians vs. larger traffic, but this is not itself a pedestrian-specific source, so marked (FAQ, RR3, extrapolated).

6. Cops

• No clean on-bike cop collision was found in the PC1/PC2 batches reviewed (cop rider "Mike" appears as a HUD name/approaching figure in pc1_J_oppcrash.jpg and pc1_J2_guardrail.jpg but no impact resolves within those sheets). (clearly unanswerable from the current frame set without further extraction)
• RR3 FAQ gives a detailed, if Genesis-RR3-specific, breakdown of cop contact that is the best source found on this:
  • Static "road block" cop cars that just idle in the road do not bust you on contact: "If you hit them, they won't arrest you." (RR3 FAQ L104-106)
  • Motorcycle cops, cop cars, and cop helicopters are the three kinds that do arrest on contact/proximity; hitting a blocking cop car is "an instant bust... Probably the most lethal cops", and the helicopter "will hover above you, then smash down on you. Instant bust." (RR3 FAQ L108-126)
  • So, in this game's design at least, cop contact severity is binary (bust/no bust) rather than a graded damage event like hitting a car or rider. Marked (FAQ, RR3) — not confirmed for the PC original, where road-rash-research.md documents the "Busted" state (stamina/time-out via stopping) but not a contact-triggered bust specifically.

7. Fallen rider / fallen bike as an obstacle

• A downed rival and a separated, lying bike are themselves visible as static hazards in the lane for the following several seconds, not instantly cleared: confirmed across pc1_A_after.jpg (downed rival passed by three more HUD-named racers in turn — Bose, then Jim — before the sheet ends) and pc1_I_fallenbike.jpg (a fallen motorcycle with a red-helmeted rider beside it, stationary at the roadside). (observed in video)
• Whether riding into a fallen rider/bike causes a crash of your own, as opposed to merely being a visual hazard to dodge, was not resolved in the reviewed frames — no sheet shows the player actually making contact with one. (clearly unanswerable from the current frame set without a targeted extraction aimed at a near-miss/hit on a downed rider)

8. Roadside objects (tree, pole, sign, building, guardrail, cliff)

• road-rash-research.md §2.5 already documents the general rule from text sources: fixed obstacles are sorted per track into things that launch you (jump ramps) and things that crash you, and the RR3 FAQ's animal-tier breakdown (§5 above) shows the same sliding/small-jump/crash three-way split applies to small objects as well as animals. Video does not add a new fact beyond this for named roadside objects specifically: pc1_J2_guardrail.jpg shows the player riding alongside a guardrail-lined road without a confirmed direct impact in the frames reviewed, and trees/signs/buildings appear only as background scenery or as jump jumps described already in road-rash-research.md, not as a resolved impact in any sheet reviewed here. (clearly unanswerable for a video-sourced impact on this particular batch; the text-sourced general rule stands as the best current answer)
• Cliffs specifically: not captured — the coastal/cliff track (pc1_C_ped.jpg) only shows the player stopped near a pedestrian, not a cliff-edge departure. (clearly unanswerable from the current frame set)

9. Landing a jump

• Not resolved: no sheet in the reviewed batch shows a clear ramp-style launch and landing from current height/shadow cues; pc1_E_sierra_punch.jpg and pc1_F_oppcrash.jpg end on approach to hazards without a confirmed airborne moment. (clearly unanswerable from the current frame set without a fresh extraction aimed specifically at one of the named small-jump obstacles from road-rash-research.md §2.5, e.g. a bush/rock/pedestrian on a track known to use them as ramps)

10. Supplementary questions

• Can bikes ever pass through each other? No instance of visual clipping/ pass-through was seen in any reviewed sheet; every rider-rider approach either resolves into one bike shoving the other sideways (§1) or a crash. Marked (observed in video, negative result) rather than a confirmed "never", since the reviewed sample is a few dozen seconds out of over an hour of footage.
• Knocked into oncoming traffic: see §4 — the manual confirms this is an intended hazard reachable via the kick move, but neither manual nor reviewed video frames show the resulting second impact. (unverified)
• How cars react when hit: no reviewed frame shows a hit car swerving, stopping, or spinning out — the truck in the Napa sequence (§3) continues in its lane in the frames immediately after the player's crash, suggesting traffic is not knocked around by the player the way riders are, but the sample is small. (observed in video, low confidence)
• What the AI does when blocked: road-rash-research.md §8.4 already covers the rubber-banding distance-keeping behaviour; no reviewed frame shows an AI rider specifically blocked by wreckage and rerouting, so nothing further to add here. (clearly unanswerable from the current frame set)

Ledger

Open / unanswerable with current evidence

• Exact MPH before/after a car-crash impact (needle not legible at current contact-sheet resolution; would need a fresh crop-and-zoom extraction on the Napa truck-crash seconds specifically).
• Player-rear-ends-rider and player-is-rear-ended-by-rider as clean, isolated events (only side-rub and car-rear-end were found).
• Oncoming-car impact, cliff departure, named roadside object (tree/pole/sign/ building/guardrail) direct impact, jump landing, and contact with a fallen rider/bike — none were captured in the batches reviewed; each would need a fresh, targeted ffmpeg extraction rather than further review of the existing rr-collide/ set, which has been exhausted for these kinds.
• What happens to the car itself and to the knocked rider at the oncoming-lane impact point (both halves of the "knocked into oncoming traffic" chain).
• Whether cop contact in the PC original (not RR3/Genesis) is a binary bust or graded, and whether a stationary roadblock-type cop exists on PC at all.