#include <bits/stdc++.h>
using i64 = long long;
constexpr i64 inf = 1ll << 50;
int calcLIS(std::vector<i64> T) {
std::vector<i64> dp(T.size() + 1, inf);
for (const auto e : T) {
auto itr = std::lower_bound(dp.begin(), dp.end(), e);
*itr = e;
}
return (int)(std::lower_bound(dp.begin(), dp.end(), inf) - dp.begin());
}
int solveNaive(std::vector<i64> T, i64 X) {
int ans = 0;
for (int i = (int)T.size() - 1; i >= 0; --i) {
T[i] += X;
ans = std::max(ans, calcLIS(T));
}
return ans;
}
int solve(std::vector<i64> T, i64 X) {
std::set<std::pair<i64, int>> dp1, dp2;
dp1.insert({0, 0});
dp2.insert({0, 0});
for (const auto v : T) {
// normal
auto itr = dp2.lower_bound({v, 0});
--itr;
int l = itr->second;
dp2.insert({v, l + 1});
itr = dp2.upper_bound({v, l + 1});
while (itr != dp2.end() and itr->second <= l + 1) {
itr = dp2.erase(itr);
}
// special
itr = dp1.lower_bound({v + X, 0});
std::pair<i64, int> p = {-1, -1};
--itr;
p = std::make_pair(v, itr->second + 1);
itr = dp2.lower_bound({v, 1 << 30});
--itr;
if (itr->second < p.second) {
dp2.insert(p);
itr = dp2.lower_bound(p);
--itr;
if (itr->first == v) {
dp2.erase(itr);
}
itr = dp2.upper_bound(p);
while (itr != dp2.end() and itr->second <= p.second) {
itr = dp2.erase(itr);
}
}
// normal
itr = dp1.lower_bound(std::make_pair(v, 0));
--itr;
l = itr->second;
dp1.insert({v, l + 1});
itr = dp1.upper_bound({v, l + 1});
while (itr != dp1.end() and itr->second <= l + 1) {
itr = dp1.erase(itr);
}
}
return dp2.rbegin()->second;
}
int main() {
int N;
i64 X;
std::cin >> N >> X;
std::vector<i64> T(N);
for (auto &e : T) {
std::cin >> e;
}
std::cout << solve(T, X) << std::endl;
/*
std::mt19937 mt(100);
std::uniform_int_distribution<i64> dist(1, 10000000);
int cases = 10000;
while (cases--) {
int N = 100;
std::vector<i64> T(N);
for (auto &e : T) e = dist(mt);
i64 X = dist(mt);
if (solve(T, X) != solveNaive(T, X)) {
std::cout << N << " " << X << std::endl;
for (const auto e : T) std::cout << e << ' ';
std::cout << std::endl;
}
}
*/
}
